The following section presents the results of the final demonstration which were validated during the final demonstration test campaign executed remotely across 4 locations (DFKI, PIAP-Space, AVS, LAAS-CNRS).
Cooperative Mapping: This cooperative mapping demo involved both robots at different locations connected by a common network for the multi-agent task planned to schedule and issue goals to each other for cooperative exploration and mapping. The agents were able to receive and plan traverses for their respective mapping goals correctly and plan traverses in the allocated area using the Rover Guidance (RG). The local/rover DEMs (Digital Elevation Models) and global DEMs (merged) for each robot were generated and provided representative maps of the test areas.
Grading and Tool exchange: For the tool exchange task, initial tests consisted of commanding the robot arm to defined pre-docking poses of the HOTDOCK interface of the gripper and shovel. Following this, the robot arm motion planner was used to command the arm to the tool interfaces pre-docking poses and perform a straight line trajectory for the docking phase. The HOTDOCK interface was commanded to lock/unlock via the Veles control center automatically using a state machine which matched the goal when within range of the alignment sensors. The grading operation was validated first via teleoperation to adjust the height of the grader blade and check its interaction with the sandy surface. The next step was to perform the grading test using RG (Rover Guidance) with waypoints, perception and localisation which was verified successfully.
Cooperative Manipulation for unloading, transport & assembling modules: The cooperative manipulation demonstration consisted in having the two robots, Mantis and Veles, using their arms to manipulate an object while their mobile bases were static. The grasping capability on the Veles manipulator was demonstrated by grasping a box (with a marker) and a handle, and moving it between predefined positions. On the Mantis a similar approach was used, but the 5 Dof manipulator with limited workspace reachability, required that the mockup with the handle be positioned within the accessible workspace via repeated trials. The physical cooperative manipulation scenario had to be further adapted for the demo as the Mantis and Veles couldn’t be tested in the same location due to COVID-19. The test was moved to a different setup which consisted of 2 fixed robot manipulators with external perturbations added manually to simulate one of the manipulators to be on a mobile base. Numerous pick and place tasks could be successfully run on the LAAS setup, showing the ability of CM2P to plan motions for a closed kinematic chain, and the ability of CM2C to control the execution of the arms motions in presence of perturbations. A virtual cooperative manipulation demo focused on the synchronous planning and motion control for executing each robot’s task, to detect the marker on an object, approach the object and grasp it virtually (grasping was demonstrated independently) and finally move it to another location.
Gantry deployment & control: The initial scenario of the ganty deployment over uneven terrain was planned using the Veles assisting the deployment by pulling one of the legs with a constant force and velocity, while the Mantis would be anchoring one of the legs. The CM2C (Cooperative Motion Controller) would coordinate this by commanding the Gantry’s motion for this deployment together with the RWAs. Due to the updated scope of the demo where robots couldn’t be co-located due to COVID-19 travel constraints, validation of this setup was performed remotely to check the commanding of the deployment and reception of the telemetry from the client interface over a virtual network. The gantry was successfully deployed and reached the deployed-secured state with the desired velocity, and then retracted back to the retracted-secured state on the flat terrain.
