Contingent Task and Motion Planning under Uncertainty for Human–Robot Interactions

Abstract

Manipulation planning under incomplete information is a highly challenging task for mobile manipulators. Uncertainty can be resolved by robot perception modules or using human knowledge in the execution process. Human operators can also collaborate with robots for the execution of some difficult actions or as helpers in sharing the task knowledge. In this scope, a contingent-based task and motion planning is proposed taking into account robot uncertainty and human-robot interactions, resulting a tree-shaped set of geometrically feasible plans. Different sorts of geometric reasoning processes are embedded inside the planner to cope with task constraints like detecting occluding objects when a robot needs to grasp an object. The proposal has been evaluated with different challenging scenarios in simulation and a real environment.

Hybrid conditional planning for robotic applications

“…In recent studies (Akbari et al, 2020; Rizwan et al, 2020), we observe the uses of hybrid conditional planning in robotics in the spirit of our proposed method. Akbari et al (2020) embedded geometric reasoning with the conditional planner Contingent-FF for human–robot collaborative manipulation tasks that involve actuation actions (e.g., push an object, open a container) and sensing actions (e.g., sense the pose of an object, check whether a container is open).…”

“…In recent studies (Akbari et al, 2020; Rizwan et al, 2020), we observe the uses of hybrid conditional planning in robotics in the spirit of our proposed method. Akbari et al (2020) embedded geometric reasoning with the conditional planner Contingent-FF for human–robot collaborative manipulation tasks that involve actuation actions (e.g., push an object, open a container) and sensing actions (e.g., sense the pose of an object, check whether a container is open). Rizwan et al (2020) used the hybrid conditional planner HCP-ASP for human–robot collaborative assembly planning that involves three types of actions: actuation actions (e.g., pick and place an object), sensing actions (e.g., sense that the human is holding a part), and communicative actions (e.g., ask human for help, offer human help, confirm some tasks).…”

Visually Grounded Task and Motion Planning for Mobile Manipulation

“…While most TAMP methods assume a fully observable and deterministic world [2], some have been developed to account for the uncertainty from perception and action outcomes [31]- [36]. For instance, the work of Kaelbling and Lozano-Pérez extended the "hierarchical planning in the now" approach to address both current-state uncertainty and future-state uncertainty [31].…”

Special Issue on Multi-Robot Systems: Challenges, Trends, and Applications

“…The paper proposes a novel method for information sharing that works when robots have different sensors, there is positional uncertainty, and obstacles are dynamic. Aliakbar Akbari, Mohammed Diab, and Jan Rosell [8] also focus on uncertainty but in the context of mobile manipulation. In these applications, humans can collaborate with robots to execute complex actions, sharing their knowledge about the task and scenario.…”