Stefan-Octavian Bezrucav, Research Assistant in robotics and mechatronics at the Institute of Mechanism Theory, Machine Dynamics and Robotics of RWTH Aachen University, has presented the paper “Improved AI planning for cooperating teams of humans and robots” during the Planning and Robotics Workshop organized during the ICAPS 2020 conference.
The paper, which results are part of Sharework project research activities, presents a new modelling approach for task planning and scheduling to accommodate the requirements from scenarios involving teams of humans and robots. The approach, named Three-Level Planning, is based on automatic planning strategies and is constructed around the ROSPlan framework.
The approach consists of a goals clustering level, an automated planning level and a hardware-close level for which state-machines containing basic actions and recovery procedures are implemented. Beside this approach, the parallel planning and dispatching features were developed, through which the waiting times that occur due to replan requests are minimised. The presented method enhances the planning system with a set of new features, filling the research gap that did not allow a realistic planning procedure for teams of cooperating humans and robots.
The Planning and Robotics workshop aimed at constituting a stable, long-term forum where researchers from both the P&S and the Robotics communities can openly discuss about relevant issues, research and development progress, future directions and open challenges related to P&S when applied to Robotics.
AI Planning & Scheduling (P&S) methods are crucial to enable intelligent robots to perform autonomous, flexible, and interactive behaviors, but a deep integration into robotic architectures is needed for their effective deployment. This requires a close collaboration between researchers from both the AI and the Robotics communities. In this direction.