Practical autonomous and semi-autonomous robots need automated plan generation if they are to achieve even simple missions with minimal human intervention. Such plans need to be communicated interactively, simply and quickly to human supervisors in a way that allows them to rapidly assess performance and risk. This is especially the case for mixed robot/human teams.
There is, therefore, an emerging interest both from the planning and visualization communities at researching real-time interactive monitoring, selection and modification of plans for robot systems and mixed teams. Such interfaces should allow: (i) quick, intuitive exploration of the space of alternative plans to allow rapid and informed selection according to different metrics (e.g. efficiency vs. risk); (ii) the easy integration of supervisor knowledge and expertise (e.g. knowledge that a particular obstructions or that battery usage policies), (iii) allow the user to accept only part of a plan and suggest modifications (rather than accepting or rejecting plans as a whole); the addition of new priorities and goals mid-plan.
This project would develop and assess an interactive, visual, browser-based interface and planner that allows humans to rapidly compare and evaluate multiple, multi-agent plans, and explore ‘what-if’ scenarios, and to modify plans and assess risk. Assessment would be via user studies. The emphasis of this project would be on the developing effective, mainly visual and highly interactive interfaces (perhaps linked with simple voice input or other modalities) with existing multi-agent, temporal planners for real world situations.
Freedman, R.G., et al. “User Interfaces and Scheduling and Planning: Workshop Summary and Proposed Challenges”, 2018 AAAI Spring Symposium Series.
Planning tools: http://planning.domains/
ROSPlan Framework: https://github.com/KCL-Planning/ROSPlan/wiki