Our second objective is to develop algorithms for the physical simulation of humans interacting with robots. Here the challenge is to be able to simulate their mutual interaction and in particular the effect that the robot has on the human body, considering individual factors and aligning as much as possible the simulation to the reality.
Toward realistic simulation of humans and their tasks
Our ambition is to reduce the reality gap in the simulation, making sure the simulation is coherent with human physics and biomechanics. In the line of our past work, we seek a trade-off between the reality gap and the computational complexity by developing rigid body models and using Quadratic Programming approaches and Model Predictive Control. We aim to have a realistic physical simulation in terms of whole-body kinematics and dynamics features and to obtain joint torques and efforts that are coherent with physiological measures.
For the next years, HUCEBOT aims at benefiting a more realistic simulation of the human body and its movements. To reach such a result, we take interest in four features: the rigid-body model, simulation of interactions with wearable devices, models of fatigues and the diversity and variability which characterize human actions. The overall goal of benefiting from better models enable our team to improve a quality of the simulations and algorithms for decision and assistance as well as to simulate real-world tasks more faithfully. Therefore, robots could move faster, in a more realistic way, interact with objects and manipulate heavy payloads.
Sensing and software
Adequate sensing is necessary to compare the simulated data with the ground truth. Our team conducts lab studies to gather data about physiological status (posture, muscle activation, cardiac activity, contact forces and center of pressure) thanks to various wearable sensors including IMUs, motion capture suits, EMG, EKG and force plates. The gathered data is then used for training machine learning models and impacting the evolution of the human simulation.
Developing suitable software tools that provide ergonomics and motion estimations for active control in real-time and in the context of human-robot interaction is a clear challenge. In the last years, we have developed several software tools for digital human model simulation, ergonomics estimation and visualization. We will continue this software effort, since we aim, in the long term, at a unified software for ergonomics and dynamics simulations of humans that will be a powerful tool and a building block for our team’s research on human-centered robotics, and that could also be shared to our collaborators outside the team.
