CASTING ROBOTIC END-EFFECTORS TO REACH FAR OBJECTS IN SPACE AND PLANETARY MISSIONS

In several space robotics applications as well as in planeta ry xploration missions, the possibility of reaching large workspace would afford great potential advantages. To oper ate on objects at distances several times larger than the physical dimensions of the robot, mobile platforms [1] equi pped with articulated arms are practically the only availab le solution at the state of the art. However, wheeled or legged r obotic locomotion depends heavily on the characteristics of the terrain, and is usually forced to trade velocity of exe cution for robustness to terrain asperities. As a matter of fact, e.g., Martian explorers Spirit and Opportunity have b e n traveling at max. 180 m/h speed, on an average mission length of 100 meters from the base station, thus limiting the number of samples returned per day. The alternative of building arms with either very long links [2] or many links [3 ], [4] seems to be applicable only in some very specific cases – for instance in the absence of gravity – and yet impose s the use of very wide mechanical structures despite the extension of their reachable spaces. In this paper we present work aimed at developing a compact ro botic device able to reach objects at far distance. The work is based on the idea of casting manipulation, a robotic technique that was proposed in [5], and that allow s to deploy an end-effector at large distances from the robot’s b ase by throwing (casting) it and controlling its ballistic fl ight using forces transmitted through a light tether connected t o the end-effector itself. The tether cable of the robotic de vice can also be used to retrieve the end-effector, and to exert fo rces on the robot’s environment. The operating phases of casting manipulation comprise a startup phase, a steering phase, and anobject-return phase. During the startup phase, the robot is controlled so as to impart the end-effector suffi cient mechanical energy to reach the target-object. When the first phase concludes, the end-effector is thrown and its trajectory is steered by means of forces transmitted throug h the tether cable in order to approach the moving object with s uitable orientation and velocity (steering phase). Once the object has been caught, the tether cable is reeled up and t he object is retrieved (object-return phase). Fig. 1 depict s possible application of casting manipulation during sampl e-return missions (left), and outlines the different opera ting phases of the technique (right). Ability of simple casting manipulator prototypes to fetch f araway fixed objects, controlling the position and orientation of the gripper and even avoiding obstacles, has been d monstrated in [6]–[8]. More recent work has extended applicability of the technique to reach targets with uncert ain position, or that are possibly moving [9], by using simplified yet accurate models that are suitable for realtim e computation, and visual feedback of the moving targets. Starting from that result, we will describe two control sche mes of the steering phase: the first approach is simpler and