Can you predict human movement to help robots work safely alongside people?
As robots move out of safety cages and into warehouses, factories and homes, they need to anticipate what nearby people will do next. In this global challenge, you will build models that forecast human motion and collision risk using real-world human-robot interaction data from the open source CHICO Pose Forecasting, supplemented with new synthetic scenarios. AMD-sponsored cloud compute and robotics learning resources will be available to all participants. The top 40 teams in the first phase will enter the second phase of the challenge, where they submit their code to be tested against unseen human-robot scenarios in a simulation environment built on Gazebo, MuJoCo or Genesis. This challenge is open to all, sponsored by AMD and powered by Zindi.