Why Touch Is Essential to the Future of Physical AI
Haply co-founder Colin Gallacher joined Robot Builders Club host Ali Afzal for an in-depth conversation about haptics, robotics, Physical AI and the future of human-machine interaction. Their discussion explores how force-feedback technology can help people control robots, demonstrate physical tasks and communicate human knowledge to intelligent machines through motion.
Published July 25, 2026
Key takeaways
Traditional computer interfaces were designed for two-dimensional screens, not physical interaction in three-dimensional environments.
Force-feedback haptics can help users feel contact, resistance and movement while controlling robots or interacting with digital objects.
Motion can function as a language for communicating physical knowledge to robots.
Simulation and synthetic data are valuable, but real-world human demonstrations remain essential for refining robotic behaviour.
Teleoperation and shared autonomy will continue to matter as robots encounter unfamiliar tasks and unpredictable environments.
Haply’s immediate focus includes robotics data collection and policy training, alongside professional 3D modelling and design workflows.
Accessible interfaces and well-designed workflows can help more people control and train robots without specialized robotics expertise.
Moving beyond the mouse
Human hands allow us to navigate the physical world almost effortlessly. We can pick up unfamiliar objects, adjust our grip, sense resistance and respond to changes in weight or position without consciously calculating every movement.
Replicating that ability through a conventional computer interface is difficult.
A mouse was built to point, click and navigate a two-dimensional screen. It cannot naturally communicate the movement, orientation and force involved in controlling a robot arm, manipulating a digital model or interacting with an object in three-dimensional space.
In his conversation with Ali Afzal, Colin described Haply’s technology as a type of three-dimensional mouse.
Instead of moving a cursor across a flat surface, a haptic device allows a person to move naturally through space while receiving force feedback from a robotic or digital environment.
That feedback can communicate contact, resistance, boundaries and physical constraints. It gives the user more information than vision alone and creates a more direct connection between human movement and robotic or digital action.
This becomes especially important when controlling robots in hazardous industrial environments, performing remote tasks underwater, collecting scientific samples or manipulating small and delicate components.
Why force feedback matters
The term haptics covers several forms of touch-based feedback. The vibration from a mobile phone is one familiar example.
Haply focuses on force feedback.
Rather than only producing a vibration, force-feedback technology can apply resistance or movement in response to what is happening inside a robotic or digital environment. A user can feel when a virtual tool reaches a surface, when a robot encounters an obstacle or when additional force is being applied.
This physical information can improve precision because the user is not relying entirely on what they can see.
The concept has already demonstrated its value in fields such as surgical robotics. Surgeons can control sophisticated robotic systems through interfaces that translate natural hand movements into precise robotic actions without requiring them to think like robotics engineers.
Haply is working to bring that same level of intuitive control to a broader range of applications, including industrial robotics, teleoperation, research, simulation and 3D design.
The same fundamental interface can support tasks across very different physical scales. These may range from manipulating electronic components smaller than a grain of sand to controlling heavy equipment in remote mines, underwater environments or other inaccessible locations.
Designing technology people can trust
For haptic technology to become widely adopted, performance alone is not enough.
Because a force-feedback device can actively move and apply forces, it must behave predictably. If the device shakes unexpectedly, moves across the desk or responds in a way the user does not understand, trust can disappear immediately.
Colin explained that this makes user experience one of the most important and difficult parts of product development.
A well-designed haptic interface should feel natural, stable and understandable. Users should not have to wonder whether they are operating it incorrectly or whether the device will respond unexpectedly.
This requires more than precise hardware. It requires thoughtful software, safety systems and interaction design.
Every behaviour must be considered. How should the user feel when a digital object snaps into position? What should happen when a tool reaches the boundary of its workspace? How should the device communicate a surface, collision or change in resistance?
The best safety and usability features are often the ones the user never notices. They simply make the experience feel reliable.
For Haply, this focus on trust and accessibility is central to building technology that can move beyond specialized laboratories and into everyday professional workflows.
Motion as a language for Physical AI
One of the strongest ideas in Colin and Ali’s conversation is that motion can function as a language.
Text and speech allow people to communicate ideas to language models. We provide instructions, review the result and refine the output until it reflects what we intended.
Physical AI requires a similar process, but the information being communicated is movement.
A person needs a way to demonstrate how an object should be picked up, how much force should be applied, how a tool should move through space or how a robot should respond when something unexpected occurs.
Haptic interfaces can help capture that motion precisely.
Just as keyboards and speech-to-text systems provide an interface for language-based AI, haptic devices can provide an interface for communicating physical knowledge to robots.
This creates a workflow for demonstration, evaluation and refinement. A person can show a robot how to perform a task, review the resulting behaviour and provide additional demonstrations where the system needs improvement.
In this context, motion is not simply movement. It is information.
Combining simulation with real-world demonstrations
Simulation and reinforcement learning can generate large amounts of synthetic training data for Physical AI.
These tools are valuable because they allow robotic systems to practise tasks repeatedly without requiring continuous access to physical hardware or real-world environments.
However, simulation alone may not fully represent the complexity of the physical world.
Materials differ in hardness, weight, texture and flexibility. Objects can move unexpectedly. Tools can slip. Environments can contain damage, obstructions and other conditions that are difficult to reproduce perfectly.
Real-world human demonstrations help provide the context that simulation may miss.
Colin compared this challenge to training a language model only on content generated by another language model. Synthetic information can be useful, but it still benefits from grounding in authentic human knowledge and behaviour.
For robotics, human demonstrations can provide that grounding.
A person can demonstrate the full task or focus only on the section where the robot is struggling. Those demonstrations can then be used to refine a policy without retraining the entire behaviour from the beginning.
Simulation and human input are therefore complementary. Simulation supports scale, while real-world demonstrations help improve precision and relevance.
Why robots will still need human guidance
As robotics advances, it is tempting to assume that full autonomy will eliminate the need for human control.
Colin expects the transition to be more gradual.
Robots may become highly capable at completing common or repetitive tasks, but the physical world contains an enormous number of edge cases. An unfamiliar door handle, a damaged component, an unusual object or an unexpected obstruction may require human judgement.
Teleoperation allows a person to step in when the robot reaches one of those situations.
The operator can resolve the problem remotely, and that interaction can become new training data for future versions of the system. Over time, the robot may learn to handle more situations independently.
This creates a model of shared autonomy in which people supervise multiple robots and intervene only when needed.
The number of robots may grow substantially, even as each operator controls them less frequently. Instead of one person continuously operating one machine, a person may oversee a fleet and provide support during difficult moments.
Colin described teleoperation as an important part of this decade’s robotics development, with autonomy increasing gradually as systems collect more data and learn from real-world interventions.
Workflows will determine adoption
Better robots and more powerful AI models are only part of the solution.
People also need practical ways to use them.
Throughout the interview, Colin repeatedly emphasized the importance of workflows. A system may be technically advanced, but it will not create meaningful value if people cannot understand how to demonstrate a task, correct an error, review the resulting data or deploy the robot in a real working environment.
These workflows must also be accessible to people who are not robotics integrators.
A warehouse operator, researcher, designer or technician should be able to work with a robotic system without needing to understand every underlying calculation, coordinate system or control loop.
This is one of the reasons Haply is focused on creating intuitive interfaces that translate natural human movement into robotic action.
The easier it becomes to demonstrate, refine and repeat physical tasks, the more effectively organizations can collect data and develop useful Physical AI applications.
Haply’s current focus
Haply’s immediate commercial focus sits across two core areas.
The first is robotics data collection, teleoperation and policy training. Haptic interfaces can help people demonstrate precise physical tasks, control robotic systems and generate the real-world motion data needed to improve Physical AI.
The second is professional 3D modelling and design.
These applications may appear different, but they are connected by the same underlying capability: allowing people to describe motion and interact naturally in three dimensions.
At CES 2026, Haply received two innovation awards, recognizing applications in robotics data acquisition and 3D design.
Haply’s partnership with Hexagon further reflects the connection between haptic interaction and professional 3D design workflows, including work with Geomagic Freeform.
Together, these areas demonstrate how the same haptic interface can support both physical machines and digital environments.
From robotics to 3D design
Haptic technology has long been used in areas such as digital sculpting and 3D modelling.
It can allow designers to feel virtual surfaces, position objects spatially and manipulate digital content more directly than with a traditional mouse.
During the conversation, Colin and Ali discussed professional workflows across software and development environments such as Geomagic Freeform, Blender, Maya, Unity and Unreal Engine.
For designers, artists, educators and developers, haptics can create a more natural way to interact with three-dimensional content.
MinVerse is part of Haply’s effort to make that interaction more accessible. It can support familiar desktop interaction and transition into a three-dimensional force-feedback interface, helping users move between established workflows and new forms of spatial control.
In the longer term, similar interfaces could support gaming, interactive learning and other consumer applications. Haply’s present focus, however, remains on professional and productivity-oriented workflows where haptics can provide clear value.
Building a platform for human-robot collaboration
Haply’s technology is designed to support different levels of precision, accessibility and application complexity.
Inverse3 provides precise force-feedback control for robotics, simulation, research and advanced technical workflows.
MinVerse brings intuitive three-dimensional interaction to design, education, creative work and other accessible professional applications.
HARP, the Human Advanced Robotics Platform, helps users connect Haply interfaces with robotic systems, establish teleoperation workflows and support the collection of data for Physical AI development.
Together, these technologies provide a foundation for interacting with both digital and physical systems through movement and touch.
The goal is not simply to create another controller. It is to build a platform that helps people communicate physical knowledge to intelligent machines.
Making Physical AI more human
The future of robotics will depend on more than autonomy.
It will depend on how effectively people can teach robots, supervise their work and intervene when the physical world presents something unexpected.
Touch provides a valuable part of that connection.
Force-feedback technology can help people understand how a robot is interacting with its environment, demonstrate how a task should be performed and interact with digital or physical environments more naturally.
As Colin and Ali discuss throughout the interview, the challenge is not only to build more capable machines. It is to create the interfaces and workflows that allow people to collaborate with them.
By making physical interaction more precise, accessible and intuitive, haptics can help ensure that the future of Physical AI remains connected to the people guiding it.
Watch the full conversation between Haply co-founder Colin Gallacher and Ali Afzal on the Robot Builders Club podcast to learn more about haptics, robot training, teleoperation and the future of human-machine interaction.
