Category: Editorial Articles

Written by Dr. Katia Bourahmoune, Acting Co-Lead, Quality Assurance & Compliance program. 

In April 2026, a humanoid robot crossed the finish line of a Beijing half-marathon in fifty minutes and twenty-six seconds, faster than any human being has ever run that distance [1]. Months earlier, humanoid robots had performed incredibly complex, synchronized martial arts routines [2]. As the media coverage gained widespread attention, something more interesting than this engineering achievement emerged: a question, not yet fully formed, about what kind of world we are now entering, and whether we are entering it with our eyes open. 

The choice to build robots in the human form is sometimes caricatured as a vanity of engineers or a concession to popular culture and science-fiction media, however, its philosophical wager is of considerable depth. The world into which these machines are being released (its factories, hospitals, construction sites, and even homes) was designed for users that stand upright, use two hands, and react dynamically to the world around them. 

Traditional forms of industrial and collaborative robots were built for tasks in bounded environments, e.g. a wheeled platform optimised for a warehouse floor or an articulated arm for a single weld point on an assembly line. A humanoid robot carries an inherent optimism about general physical intelligence: the bet, or perhaps ambition, that a machine capable of inhabiting the full texture of human environments can in time respond to the full texture of human need. The ancient concept of Ziran in classical Chinese thought illuminates what the designers in this field are reaching toward. Ziran is often rendered as naturalness, or the disposition of things to accord with their own nature [3]. In the context of robotics, this can be found in the idea of building machines that fit the world as it is, rather than demanding the world be remade to fit the machine. 

Humanoid robots are now operating in production environments and shipping in volumes that would have seemed premature as recently as 2023. Venture capital investment in humanoid robotics exceeded three billion dollars in 2024, with reports of multi-billion market projections for the next decade [4]. What this momentum cannot easily tell us is whether the design assumptions underlying this transition have been adequately examined. The present dominant commercial logic treats humanoid forms primarily as a means of fitting machine labour into existing human infrastructure i.e. same floor plan, same tools, and minimal workflow redesign. That is a reasonable engineering position. It may also be eclipsing, earlier than is wise, questions around whether the humanoid is best understood as a substitute for human presence or as a platform for augmenting it. 

It is worth noting that the world these machines are being designed to inhabit was itself built around the human body. Every dimension of that infrastructure, accumulated across two centuries of industrial development, was calibrated to the physical limits and capabilities of the biological human form. While previous waves of automation reshaped work around the machine, the humanoid, at least in aspiration, inverts that relationship. In doing so, it raises a concern that the industrial revolution never had occasion to face: What becomes of the human body’s centrality to working life when the physical form that justified building the world around it can be replicated, scaled, and indefinitely reproduced? That this question is now being asked simultaneously in boardrooms, parliaments, and papal encyclicals is perhaps the clearest measure of its weight [5].  

What the field of collaborative robotics has understood for some time (and what the humanoid moment is now forcing into general visibility) is that matching human physical capability, however necessary, is not sufficient. Harder questions concern the relationship between human and robot: what kind of human-robot partnership produces durable, humane, and useful outcomes, and under what conditions workers can reasonably extend trust to machines working beside them. Those questions shaped decades of research into human-robot interaction and collaboration and the work the Australian Cobotics Centre has been part of since 2021. How much humanoids come to define the next chapter of that work is ultimately a question research and humanity will have to answer. 

 

Call for Participation

The Australian Cobotics Centre is calling for experts across academia, industry, and government to participate in a research study at the University of Technology Sydney aimed at developing a clearer definition of collaborative robots. Participation involves an online discussion followed by a brief activity to rate statements about cobots. Your input will directly inform how the field defines and frames human–robot collaboration. 

More information and EOI here:  EOI and Consent Form (https://forms.office.com/r/YSYQPWqD8X) 

 

References and Further Reading:  

[1] Harmon, K. (2026). A humanoid robot beat the human half-marathon record at a Beijing race. But what did it actually prove? Scientific American.  

[2] Unitree Robotics. (2026). Kung fu meets spring: Unitree Spring Festival Gala robots present “Cyber Real Kung Fu” in the year of the horse [Press release]. PR Newswire.  

[3] Cleary, T. (Trans.). (1992). The essential Tao: An initiation into the heart of Taoism through the authentic Tao Te Ching and the inner teachings of Chuang-tzu. HarperCollins.  

[4] Goldman Sachs. (2024). Humanoid robots: A $38 billion market by 2035. Goldman Sachs Research.  

[5] Leo XIV. (2026). Magnifica Humanitas [Encyclical letter]. Dicastery for Communication, Holy See.  

 

 

Industrial automation has traditionally been built around a fundamental limitation: robots cannot “feel”. Sensing physical interaction with a workpiece or environment has historically required expensive hardware, making such capabilities impractical for many industrial systems. As a result, conventional industrial robots typically operate in isolation, executing preprogrammed motions without direct awareness of the forces they encounter. 

The development of collaborative robots (cobots) introduced the ability to sense internal forces and detect collisions, allowing for safer human-robot interaction. However, true physical awareness requires external sensors. When equipped with exteroceptive sensors, such as those that measure forces or vibrations, robots can respond to external conditions like a changing workpiece. This capability expands robotic automation into applications that require both force sensitivity and precision, including complex finishing operations such as grinding and polishing. 

Grinding remains one of the most physically demanding tasks in metal fabrication. The process requires a balance between force and precision; too little pressure slows production, while too much risks damaging the part or wearing down the tool prematurely. These characteristics make grinding a promising candidate for automation. Many manufacturers pursue robotic grinding not only to address rising labour costs and workforce shortages, but also to achieve process consistency and repeatability. 

However, implementing robotic grinding typically requires high-end sensing hardware. These systems often rely on force/torque sensors to measure the interaction between the tool and the workpiece, enabling robots to maintain the controlled force necessary for precision finishing. These sensors can cost upwards of $4,400 USD. For many small and medium-sized enterprises (SMEs), particularly in Australia, this cost represents a significant barrier to entry, turning automation into a financial hurdle rather than a competitive advantage. 

Recent research by PhD candidate, Zongyuan Zhang and his supervisory team suggests that robots may not need expensive sensors to achieve force awareness. Human operators performing grinding tasks often rely on subtle auditory cues, like the pitch and vibration of the tool, to judge the quality of contact with the material. Experienced machinists can detect changes in force or tool wear simply by listening to the sound of the process. Inspired by this intuition, researchers have begun exploring whether similar information can be extracted using low-cost acoustic sensing combined with machine learning. 

The proposed Acoustic Feedback Robotic Grinding (AFRG) system (see Figure 1) demonstrates how this approach can work in practice. Instead of measuring force directly, the system monitors the acoustic signature of the grinding process. A single contact microphone is mounted to the tool bracket, capturing vibrations transmitted through the structure of the tool while filtering out much of the ambient noise present on a factory floor. 

The captured signal is processed by a specialised neural network known as PSDRegNet, a two-dimensional convolutional neural network designed to estimate the grinding force. By learning the complex relationship between acoustic patterns and grinding forces, the model can estimate the interaction force in real time. This data can then be used to adjust the robot’s behaviour online. Since the system learns this relationship directly from data, it avoids the need for rigid mathematical models that would typically govern robotic finishing processes. This flexibility allows the same system to adapt more easily to different materials, tools, and process conditions, reducing the time and engineering effort required to reconfigure robotic cells for new tasks. 

Another challenge in robotic finishing is tool degradation. As grinding discs wear down or become clogged with material, their cutting efficiency declines. Robots that rely on fixed paths or constant force setpoints often struggle to compensate for this change, leading to inconsistent material removal over time. In experimental trials conducted on hardened stainless steel, a material known for accelerating tool wear, the AFRG system demonstrated a fourfold improvement in grinding consistency compared to traditional force-based control. Since the acoustic model captures tangential force information closely related to the material removal rate, the system can maintain a stable finishing process even as the physical properties of the grinding disc change. 

The implications extend beyond grinding. If meaningful process information can be extracted from inexpensive sensors such as microphones, accelerometers, or cameras, machine learning may enable a new generation of low-cost perceptual capabilities for industrial robots. Instead of relying on specialised hardware for every sensing task, robots could infer key physical variables from readily available signals. 

For the Australian Cobotics Centre, this approach demonstrates a cost-effective pathway for quickly upgrading existing industrial infrastructure, something important for many Australian SMEs. Many legacy robots are position-controlled, following a set of predefined positions without sensing the forces involved in the task. Retrofitting these systems with force sensors can be costly, but in contrast, an acoustic sensing system can be integrated with minimal modifications, offering closed-loop force control at a fraction of the cost. 

More broadly, this work challenges the assumption that precision automation must rely on expensive hardware. By combining off-the-shelf sensors with machine learning, it becomes possible to convert robots from pre-programmed machines into adaptive systems capable of responding to their environment. 

 

 

 

 

 

 

 

Acoustic Feedback for Closed-Loop Force Control in Robotic Grinding, Zongyuan Zhang*, Christopher Lehnert, Will Browne, Jonathan Roberts 

Written by: Akash Hettiarachchi, Melinda Laundon, Penny Williams and Greg Hearn, all based at QUT in the Australian Cobotics Centre’s Human‑Robot Workforce program

International Women’s Day is an opportunity to celebrate progress toward gender equity and to reflect on persistent structural challenges in the workplaces. While many sectors highlight successes in advancing gender diversity, Australian manufacturing continues to struggle with its historically male-dominated image. Gender inequality in manufacturing is widely recognised. Yet the sector often progresses with uniform policies and strategies.

Our recent research, published in Equality, Diversity and Inclusion: An International Journal, challenges this approach by revealing that diversity patterns across manufacturing are far more complex, uneven, and sub-sector specific. This study examines workforce diversity across Australian manufacturing using Australian Census data from 2006 to 2021. By analysing trends in gender, generation, ethnicity, disability and educational qualifications across manufacturing sub-sectors, we show why improving gender equity requires targeted, context-specific strategies rather than generic, sector-wide approaches.

Manufacturing Gender Diversity is Uneven and Complex

Australian manufacturing is often described as male-dominated; however, our analysis reveals significant variations in workforce gender diversity across its sub-sectors. The overall representation of women differs considerably among sub-sectors such as food and beverage manufacturing, machinery and equipment manufacturing, and fabricated metal products. This unevenness raises questions about the success of gender-specific diversity strategies and outcomes in Australian manufacturing. Given the diversity composition differences among sub-sectors, broad, blanket gender diversity strategies are unlikely to be effective. Instead, improving gender equity requires a clear understanding of where women are over-represented, under-represented, or entirely absent. It also requires an understanding of how personal, structural, and occupational patterns differ across various manufacturing sub sector contexts.

True Representation is More than Increasing Participation

One of the key findings from our study is that improving gender equity is not simply about increasing the overall number of women in manufacturing. Women are frequently concentrated in specific roles and occupational categories, with limited representation across many technical and operational jobs on the production floor, compared with administrative functions. A focus on numbers alone does not deliver sustainable or meaningful representation in most needed job roles in operations.

These patterns suggest that recruitment focused strategies, while important, are insufficient. Genuine progress requires deeper organisational attention including job design, skills development, promotion pathways, and workplace cultures that support retention and advancement for equal opportunities of all genders. Gender equity in manufacturing is therefore closely tied to how work is organised and how careers are structured, particularly as roles continue to evolve through automation and digitalisation.

Different Generations and Future Skills

Our research highlights a persistent structural challenge within Australian manufacturing: the significant representation of an ageing workforce, alongside ongoing difficulty in attracting younger workers (particularly young women) into manufacturing careers. Despite numerous government initiatives, this imbalance remains largely unchanged.

Older workers continue to play a vital role, contributing critical operational knowledge, continuity, and deep technical expertise. At the same time, long-term workforce sustainability depends on successfully attracting and integrating younger talent. Compared with other sectors, manufacturing has been less successful in renewing its workforce, creating a growing concern for future labour supply.

These demographic dynamics intersect directly with technological change. As Industry 4.0 technologies, including collaborative robots, reshape manufacturing work, new skill demands emerge, often accompanied by workforce adjustment challenges. In response, some organisations must prioritise reskilling existing employees, while others may need to rethink job design and career pathways to better align with evolving technologies and the expectations of a more diverse future workforce.

From a gender equity perspective, this underscores the importance of expanding access — not only to employment, but also to training, reskilling, and progression opportunities. Without deliberate intervention, technological transformation risks reinforcing existing gender patterns rather than enabling more inclusive manufacturing careers.

Why This Matters for Cobotics and The Future Of Work

From the perspective of the Australian Cobotics Centre’s Human‑Robot Workforce research program, these findings reinforce that workforce diversity is central to successful technology adoption. Collaborative robots are introduced into existing workplaces shaped by workforce demographics, skills and organisational practices.

Manufacturing sub‑sectors with different gender profiles and labour market conditions will experience cobot adoption in different ways. Without inclusive workforce strategies, new technologies risk reproducing existing inequalities. Conversely, when job design and skill development are approached with gender equity in mind, collaborative robotics can support safer, more sustainable and more attractive manufacturing work.

Turning Reflection into Sustained Action

International Women’s Day is a useful moment for reflection, but our research highlights the need for ongoing, evidence‑based action. Gender inequality in manufacturing is well recognised, yet it is often oversimplified. Addressing it requires sub‑sector‑specific strategies informed by data and grounded in the realities of different manufacturing contexts.

At the Australian Cobotics Centre Human-Robot Workforce Research Program, this research informs our work on future skills, job design and workforce readiness. Improving gender equity is not separate from productivity or innovation. Rather, it is integral to building a manufacturing workforce capable of adapting to technological change and supporting the long‑term sustainability of Australian industry.

 

By Munia Ahamed, UTS PhD Researcher, Australian Cobotics Centre

Technology adoption decisions in manufacturing are often characterised by a tension between perceived opportunity and perceived risk. This is particularly true for small and medium enterprises considering investments in collaborative robotics and automated quality systems, where the upfront costs are concrete, but the returns can feel uncertain.

Research consistently identifies this uncertainty as a key barrier. A 2024 UK government review of advanced technology adoption found that financial barriers—particularly difficulties justifying investment decisions due to uncertain returns—ranked among the most frequently cited obstacles for manufacturers [1], [2], [3]. Similar patterns emerge in studies of Australian SMEs, where decision-makers report hesitancy in embracing new technologies despite recognising their potential benefits.
This article argues that one way to address this challenge is through a more rigorous application of quality cost economics—a well-established body of theory that provides frameworks for quantifying the costs of defects, rework, and quality failures. By grounding technology adoption decisions in these frameworks, manufacturers can move from intuition-based decision-making toward evidence-based investment analysis.

The Economics of Quality: Theoretical Foundations
The concept of quality costing has a long history in operations management. Joseph Juran introduced the notion of the “cost of poor quality” in his 1951 Quality Control Handbook, arguing that organisations inevitably pay for quality—either through prevention and detection, or through the consequences of failure. His work established that appraisal and failure costs are typically much higher than prevention costs, suggesting that investment in getting things right the first time yields significant returns.

Philip Crosby extended this thinking in the 1970s and 1980s with his influential argument that “quality is free”. Crosby’s position was not that quality improvement carries no cost, but rather that the price of nonconformance—scrap, rework, warranty claims, lost customers—far exceeds the price of conformance. His research suggested that well-run quality programs could yield gains of 20 to 25 percent of revenues, with the cost of nonconformance reducible by half within 12 to 18 months of systematic effort.

Armand Feigenbaum’s Prevention-Appraisal-Failure (PAF) model provides a useful taxonomy for categorising quality costs. Prevention costs include activities designed to avoid defects occurring in the first place—process design, training, quality planning. Appraisal costs cover inspection, testing, and measurement activities. Failure costs, both internal (scrap, rework) and external (warranty, returns, reputation damage), represent the consequences of quality problems that were not prevented or detected.

In practical terms, this means that spending more on preventing defects upfront usually reduces the overall cost of quality problems later. Failure costs tend to decrease faster than prevention costs increase, so the total quality cost goes down. This insight remains as relevant today as when it was first articulated, and it provides a theoretical basis for evaluating investments in quality-enhancing technologies.

The Economics of Detection Timing
A related concept concerns the timing of defect detection. The 1:10:100 rule, attributed to George Labovitz and Yu Sang Chang (1992), captures the exponential escalation of costs as defects progress through the value chain. In its simplest form, the rule suggests that addressing a problem at its source costs one unit; finding and correcting it later in the process costs ten units; and dealing with it after it reaches the customer costs one hundred units.

While the specific ratios vary by context, the underlying principle is well-supported: defects that escape early detection accumulate additional processing costs, and defects that reach customers incur costs that extend beyond direct remediation to include relationship damage, complaint handling, and potential regulatory consequences.

This principle has direct relevance to technology adoption decisions. Automated inspection systems and vision-guided collaborative robots do not merely accelerate quality checking—they fundamentally alter when inspection occurs. Real-time, in-process detection catches problems before they accumulate downstream costs, shifting the organisation’s quality cost profile in favourable directions.

Barriers to Evidence-Based Decision Making
If the theoretical case for quality investment is strong, why do manufacturers—particularly SMEs—struggle to act on it? The literature identifies several contributing factors.
First, quality costs are often poorly measured. While direct costs like scrap and rework may be tracked, hidden costs—inspection time, schedule disruption, expedited shipping to replace defective goods—frequently go unrecorded. Without accurate baseline data, it becomes difficult to project returns on quality-enhancing investments.

Second, uncertainty about technology performance creates decision paralysis. Studies of SME technology adoption consistently find that decision-makers hesitate when they cannot point to demonstrated results in comparable contexts. This creates a circular problem: evidence is needed to justify investment, but evidence comes from having invested.

Third, competing priorities and resource constraints mean that quality investments must compete with other demands on limited capital. In this environment, investments with uncertain or difficult-to-quantify returns tend to be deferred in favour of more immediately tangible needs.

ROI Calculators as Analytical Tools

One response to these challenges is the development of structured ROI calculators tailored to specific technology investments. When well-designed, such tools serve several functions beyond simply generating a payback estimate.

First, they impose discipline on baseline measurement. To complete the calculator, users must quantify current defect rates, rework costs, and inspection time—data that many organisations have not systematically collected. The process of gathering this information often yields insights independent of any technology decision.

Second, they make assumptions explicit. A good ROI model does not obscure uncertainty; it surfaces it. Users can see what improvement rates are assumed, what cost factors are included, and how sensitive the conclusions are to different inputs. This transparency supports more informed discussion among stakeholders.

Third, they provide a framework for comparing alternatives. By standardising how costs and benefits are categorised, calculators enable like-for-like comparison of different technology options or implementation approaches.

The value of such tools lies not in their precision—all projections involve uncertainty—but in their capacity to structure thinking and ground decisions in operational data rather than vendor claims or general optimism.

Practical Recommendations

For manufacturers seeking to apply quality cost economics and the 1:10:100 principle to their technology decisions, several practical steps can strengthen the quality of investment analysis.
Establish quality cost baselines. Before evaluating any technology investment, spend time measuring what is currently unmeasured: rework hours, scrap rates, inspection time, defect escape rates. Even approximate figures provide a foundation for analysis that intuition cannot.

Map defect origins and detection points. Understanding where in the process problems arise—and where they are currently caught—identifies the opportunities for earlier detection. The gap between origin and detection represents accumulated cost that prevention or earlier inspection could avoid.
Use sensitivity analysis. Rather than seeking a single ROI figure, explore how conclusions change under different assumptions. What defect reduction would be needed for the investment to break even? How does the payback period shift if improvement is 20% less than projected? This approach acknowledges uncertainty while still supporting decision-making.

Consider pilot implementations. Where full-scale investment feels premature, smaller-scale trials with defined metrics can generate context-specific evidence. This reduces risk while building organisational capability and confidence.

The Path Forward
The theoretical foundations for quality cost analysis are well-established, with decades of research supporting the economic logic of prevention over detection and early detection over late. What is often lacking is the practical application of these frameworks to specific technology adoption decisions.
ROI calculators, when grounded in quality economics and used as analytical tools rather than sales devices, can help bridge this gap. They provide a structured means of translating established theory into operational decision-making, replacing intuition with evidence and making the case for investment in terms that resonate with resource-constrained decision-makers.

For Australian manufacturing to remain globally competitive, we need to accelerate thoughtful adoption of collaborative robotics and quality automation. Fact-based decision tools are one contribution toward that goal.

We welcome discussion on this topic. How has your organisation approached the challenge of justifying technology investments? What frameworks or tools have proven useful?

References
[1] Make UK and RSM UK, Investment Monitor 2024: Using Data to Drive Manufacturing Productivity. London, UK: Make UK, 2024.
[2] Make UK and BDO LLP, Manufacturing Outlook: 2024 Quarter 4. London, UK: Make UK, 2024.
[3] UK Government, Invest 2035: The UK’s Modern Industrial Strategy — Green Paper. London, UK: HM Government, Oct. 2024.

 

Learning from Demonstration (LfD) enables robots to acquire manipulation skills by observing human actions. However, existing methods often face challenges such as high computational cost, limited generalizability, and a loss of key interaction details.

This study presents a compact representation for interaction recognition in LfD that encodes human–object interactions using 2D wrist trajectories and 3D object poses. A lightweight extraction pipeline combines MediaPipe-based wrist tracking with FoundationPose-based 6-DoF object estimation to obtain these trajectories directly from RGB-D video without specialized sensors or heavy preprocessing. Experiments on the GRAB and FPHA datasets show that the representation effectively captures task-relevant interactions, achieving 94.6% accuracy on GRAB and 96.0% on FPHA with well-calibrated probability predictions.

Both Bidirectional Long Short-Term Memory (Bi-LSTM) with attention and Transformer architectures deliver consistent performance, confirming robustness and generalizability. The method achieves sub-second inference, a memory footprint under 1 GB, and reliable operation on both GPU and CPU platforms, enabling deployment on edge devices such as NVIDIA Jetson. By bridging pose-based and object-centric paradigms, this approach offers a compact and efficient foundation for scalable robot learning while preserving essential spatiotemporal dynamics.

Written by Katia Bourahmoune, UTS & Acting Co-Lead Quality Assurance and Compliance program

Heidegger describes an equipment as ready-to-hand when it disappears into practice, when its use is so seamlessly integrated that it ceases to be an object of thought and becomes instead a transparent extension of action. A hammer is not noticed as a hammer when it drives a nail effectively; it is only when it splinters or slips that it becomes it becomes an object of scrutiny, unready-to-hand, with its use questioned. In modern manufacturing, collaborative robots (cobots) occupy an uneasy position between these two states. They promise repeatability, precision, and tireless monitoring, yet they are undeniably still machines to be supervised, audited, and monitored. In compliance and quality assurance, this human oversight of machines is necessary. Afterall, compliance remains the most human part of the hyper-mechanised modern manufacturing process. This is particularly evident in heavily regulated industries like medical device manufacturing, aviation and defence, where errors are measured not only in costs but in lives and national security.

For cobots to become ready-to-hand, they must be genuinely collaborative: partners in the task rather than peripheral machinery. While collaboration in the context of human-robot interaction is hard to define and evolves as the field advances, it is useful to frame it within the level on interaction between a human and a robot. These levels range from co-existence (shared space, individual actions) to co-operation (shared space, human-guided actions), to collaboration (shared space, joint bi-directional actions). Collaboration through this lens implies shared situational awareness, legible intent, and adaptive action: the robot exposes what it “perceives” (vision, force,…), why it is acting (constraints, goals,…), and how humans can adapt, override, or teach. Such interfaces must preserve human agency and skilled technique while reducing ergonomic and cognitive load. In practice, this means adaptive assistance that yields to expert touch, explanations of proposed actions, and workflows that keep responsibility distributed rather than displaced. When collaboration works this way, it does more than improve throughput; it establishes the preconditions for assurance to be intrinsic rather than supervisory. On this foundation, compliance becomes by design: assurance embedded in action, rather than appended after it. Cobots can inspect as they assemble, verify as they position, and generate audit-grade evidence as a by-product of normal operation. Cobots can extend human judgment through continuous monitoring, allowing human inspectors to concentrate on exceptions, interpretation, and continuous improvement.

This human-robot collaboration fundamentally hinges on trust. In production, workers must believe that a cobot will act predictably and safely; in quality assurance, they must also believe that the cobot’s monitoring and record-keeping are accurate and transparent. Research on automation psychology shows the dangers of both extremes: over-trust leads to blind reliance, while under-trust leads to redundancy and disuse. The literature points to several ways for calibrated trust including reliable and predictable performance, timely feedback, options for human override, transparent explanations of decisions, and auditable records tied to actions, and here we emphasise the compliance-critical elements of legibility, traceability, and contestability. Trust, then, is not an abstract sentiment but a design commitment: when cobots make their intentions legible and their decisions contestable, human operators retain meaningful agency in the loop. This keeps human judgment engaged precisely where it adds the most value. In regulated settings, this turns assurance into a shared practice rather than a supervisory afterthought, and it reorients collaboration toward preserving and amplifying human skill rather than displacing it.

Concerns are often raised that automation “deskills” human labour, relegating workers to passive supervision. Cobots designed for compliance offer the opposite prospect. By taking on repetitive inspection tasks, cobots free human expertise for higher-order judgment: interpreting anomalies, adapting processes, and innovating in response to unforeseen conditions. The skill does not vanish; it is re-centred where it matters most. In this way, cobots not only maintain but actively sustain skill, ensuring that human judgment remains the decisive element in compliance.

The Compliance and Quality Assurance program at the Australian Cobotics Centre aims to develop practical tools that specify, monitor and evaluate human–robot collaboration using multi-modality sensing and AI for assessing compliance.

When cobots are truly ready-to-hand, i.e. useful, trustworthy, and engineered for compliance-by-design, they cease to be mere machines and become true collaborators that elevate human skill while making quality an intrinsic property of every human–robot action.

 

Further reading:  

Heidegger, M. (1962). Being and time. In J. Macquarrie, & E. Robinson, (Trans.), New York, NY: Harper & Row. 

Guertler, M., Tomidei, L., Sick, N., Carmichael, M., Paul, G., Wambsganss, A., … & Hussain, S. (2023). When is a robot a cobot? Moving beyond manufacturing and arm-based cobot manipulators. Proceedings of the Design Society, 3, 3889-3898. https://doi.org/10.1017/pds.2023.390  

Hancock, P. A., Billings, D. R., & Schaefer, K. E. (2011). A meta-analysis of factors affecting trust in human-robot interaction. Human Factors, 53(5), 517–527.  https://doi.org/10.1177/0018720811417254  

Carmichael, M. (2023). Can we Unlock the Potential of Collaborative Robots?. Australian Cobotics Centre. https://www.australiancobotics.org/articles/can-we-unlock-the-potential-of-collaborative-robots/  

 

 

When we think about robotics, especially in high, stakes contexts like surgery, we often imagine advanced machines, complex algorithms, and high, tech labs. But sometimes the best way to design the future is with cardboard, PVC pipes, and a bit of puppetry.

In our recent research, presented at DIS 2025, we explored how embodied, low, fidelity prototyping can help bridge the gap between technical development and the lived realities of people who work with robots.

The Challenge

Robotic, Assisted Surgery (RAS) is a complex and dynamic human–robot interaction (HRI) settings. Surgeons rely on tacit, embodied knowledge built up over years of practice. Engineers bring deep technical expertise. Human factors specialists understand the cognitive and ergonomic limits of people in high, stress environments.

But bringing these perspectives together can be challenging, especially early in the design process. Too often, technical development runs ahead of human needs, leaving systems misaligned with real, world practice.

Our Approach: The Kinematic Puppet

To address this, we developed the kinematic puppet: a modular, tangible prototyping tool that allows users to physically “puppeteer” a robot arm without needing code or expensive equipment.

• Built with 3D, printed joints, rotary encoders, and PVC linkages, it’s reliable, reusable, and reconfigurable.
• It integrates with virtual simulation, so physical movements can be recorded and replayed as digital twins.
• It lowers the barrier for surgeons, engineers, and designers to experiment together, making abstract ideas concrete.

Through physically roleplaying it allows participants to test scenarios and explore ideas before committing to costly development.

Putting It to the Test

We trialled the kinematic puppet in a co-design workshop focused on revision hip surgery. Surgeons, engineers, and designers gathered around a low, fidelity anatomical model, simple props, and the kinematic puppet.

Through roleplay and bodystorming, participants experimented with:

• Ergonomic tool grips (pen, style vs. drill, style).
• Spatial layouts of surgical environments.
• Cooperative control methods (e.g. axis locking, haptic boundaries).

Crucially, the kinematic puppet and surgical props helped surface tacit knowledge, things surgeons know through embodied practice but may struggle to articulate. Combined with simulation, we could capture, replay, and analyse these scenarios for further design development.

Key Takeaways

1. Embodiment matters: In robotics design, haptic feedback, posture, and movement constraints cannot be understood through software alone. Tangible tools make this knowledge accessible.
2. Hybrid methods are powerful: Combining physical roleplay with digital capture bridges creative ideation and technical precision.
3. Collaboration is essential: Designers play a key role in facilitating conversations between different stakeholders helping translate knowledge and experience between disciplines.
4. Low, fidelity ≠ low, value. Sometimes the simplest prototypes spark the richest insights.

Why This Matters Beyond Surgery

Although our case study focused on surgery, the approach has wider relevance. Any domain where humans and robots work together under constraints, manufacturing, logistics, aerospace, can benefit from accessible, embodied prototyping.

By lowering the technical threshold for participation, we can bring more voices into the design of future robotic systems. That means better alignment with real workflows, safer systems, and technology that truly supports human expertise.

A Call to Action

If you’re working in robotics, design, or any field where humans and machines collaborate:

• Prototype early, prototype tangibly. Don’t wait for polished tech, start with what people can touch, move, and play with.
• Value tacit knowledge. Invite practitioners to show you, not just tell you, how they work.
• Think hybrid. Use both physical artefacts and digital tools to capture richer insights.

The future of robotics won’t just be written in code, it will be shaped through the creative, embodied practices that help us design with people, not just for them.

We would love to hear from others: how have you used tangible or embodied methods to explore technology design in your field?

PhD Research Spotlight: Zongyuan Zhang Tackles Contact Tasks with Mobile Robots

As part of the Biomimic Cobots program within the Australian Cobotics Centre, PhD researcher Zongyuan Zhang is leading a project that addresses a key challenge in manufacturing: enabling mobile robots to perform high-precision contact tasks, such as grinding, polishing, and welding, on large, arbitrarily placed workpieces in factory environments.

Zongyuan brings a diverse background in robotics to this work. He holds an M.Sc. in Robotics from the University of Birmingham, UK, where he focused on applying deep learning to manipulator force control. His experience spans control system design, mechanical structure design, and participation in a range of innovative robotics projects—including underwater photography robots, driverless racing cars, exoskeleton mechanical arms, dual-rotor aircraft, and remote-control robotic arms—some of which are now undergoing commercialisation.

His PhD project, Contact Task Execution by Robot with Non-Rigid Fixation, investigates how robots with non-rigidly fixed chassis can maintain the accuracy, stability, and adaptability required for industrial contact tasks. These tasks typically demand hybrid force/position control and high contact forces, which are complicated by the mobility and flexibility of the robot’s base.

This research contributes to the Biomimic Cobots program’s goal of developing collaborative robots that mimic human sensing, learning, and manipulation skills. It explores:

• How a robot mimics human control to execute contact tasks like sanding and grinding.
• How augmented mobility enables task execution in large, unconstrained spaces.
• How minimal task-specific programming can be used to adapt to new workpieces and environments.

Zongyuan, based at QUT, is supervised by Professor Jonathan Roberts, Professor Will Browne, and Dr Chris Lehnert, and is working onsite at ARM Hub alongside industry partner, Vaulta. The project with industrial partners concerns the efficient and accurate removal of surface oxides from metallic materials, thereby enabling tighter bonding between metal components. This embedded collaboration ensures his research is conducted in real production environments and remains grounded in the practical needs of Australian manufacturers.

Recent milestones include the:

• Design and deployment a framework for performing industrial sanding tasks using collaborative robots.
• Utilisation of sound as a multimodal input to improve the robustness of the sanding process and enhance the cost-efficiency of the robotic system.
• Exploration of how humanoid robots can achieve high-precision performance in contact tasks.

Check out our website for the latest on his project: Project 1.1 – Contact task execution by robot with non-rigid fixation » Australian Cobotics Centre | ARC funded ITTC for Collaborative Robotics in Advanced Manufacturing