Model touch, friction, and soft contact in simulation
Named on stage by Dr. Brandon DeHart (RoboHub) at a Startup Grind Kitchener-Waterloo fireside, 16 July 2026.
What the people stuck here are trying to get done
When simulation is the core place I train artificial intelligence (AI) and test control, I want it to model touch, friction, stepping and soft interaction, so what I train in simulation transfers to the real robot.
Who needs it solved
Robotics founders and researchers who train control and AI in simulation and then can't transfer it to hardware.
What makes it hard
Sim results don't survive contact with reality, forcing expensive hardware-in-the-loop workarounds ("the robot pokes a chunk of skin or a cadaver").
Working on this?
Building toward this problem, or know someone who does? Message me on LinkedIn and the chapter will make the introduction.
Speakers can ask for a change or removal at any time: andrew@loopgrow.co.