Human-aware task and motion planning for efficient human-robot collaboration

In human-robot collaboration (HRC), the variabilityof the human behavior complicates the deploymentof robust task and motion plans, and continuous update ofplans are often required to correct the task execution. Wepresent a control-based approach to achieve robustness in theexecution despite the high time variability of human and robottasks. The proposed approach consists of two layers: taskplanning considers high-level operations without taking intoaccount their motion properties; action planning optimizes theexecution of high-level operations based on current human stateand geometric reasoning. The result is a hierarchical structurewhere the lower level provides feedback on the feasibility andthe upper level uses this feedback to (re)-optimize the processplan only when needed. The method is demonstrated in anindustrial case study where a robot and a human worker worktogether to assemble a mosaic.