Mobile robotics: autonomous exploration solution for AMR
Robotic exploration: the limits of remote operation
Preliminary stage of a mobile robot integration in an indoor environment, the mapping must allow the machine to acquire sufficient knowledge of the space in which it will evolve. The more precisely this mapping is carried out, the more efficiently and accurately the mobile robot will fulfill its tasks, whether in a tertiary or industrial environment.
Generally, this mapping is carried out by a teleoperator, who controls the robot and guides its trajectory using a controller. However, this remote operation phase has some limits:
depending on the surface area, the time spent on mapping can be important. It takes about 30 minutes to map 500 m2;
a human resource is required. The teleoperator must also be trained beforehand.
Thus, remotely operated mapping generates additional costs, that can be avoided and limited through autonomous exploration.
Mobile robot: the relevance of autonomous frontier-based exploration
To allow a mobile robotic solution to map its environment, an algorithm aimed at optimizing the robot’s trajectory has to be developed according to different autonomous exploration methodologies. The Awabot Intelligence team compared these methods to select the most relevant after studying different criteria.
The Awabot Intelligence team implemented and enhanced an autonomous frontier based exploration method, in a variety of simulation environments, then in a real environment.
In simulation, the map obtained with autonomous exploration is complete, despite a less precise obstacle resolution than on the ground truth map.
In a real environment represented by one of the Awabot Headquarter’s offices, the map obtained is complete too, but less precise than the maps obtained in simulation (expected result with real sensors)
Furthermore, experiments comparing Awabot Intelligence autonomous exploration method with teleoperators demonstrated the competitiveness of the algorithm.