Online Exploration Path Planning for a Ground Robot Using the Structure of the Environment

In deploying robots for security and safety applications, the robot should fully explore one region of the environment, such as a room, before proceeding to other regions. To achieve such a behaviour, this article proposes an enhancement of the hierarchical exploration algorithm titled ‘Technology for Autonomous Robotics Exploration’ (TARE). The algorithm uses region predictions to select appropriate parameters, with regions being user inputs or algorithm outputs. The enhancement includes a new viewpoint selection algorithm which uses a region’s boundary to filter the viewpoint choices for local path formation, ensuring full exploration of the region before moving elsewhere. The algorithm is assessed by two criteria: (1) active viewpoints’ count during exploration, and (2) cumulative wait time for viewpoints visit i.e., from identification of a viewpoint to robot’s visit to the viewpoint. Simulations were conducted in Long-T and Garage environments. In Long-T and Garage environments, the proposed algorithm’s maximum active viewpoints were 10 and 12, versus TARE’s 37 and 60. The average time to visit a viewpoint is 43 and 289 sec. for the proposed algorithm vs 294 and 448 sec. for TARE’s. Future work will evaluate manipulation actions on objects in the environment to aid exploration and coverage.