The authors introduce a framework and prospective solutions for intelligent locomotion, defined as the ability for a mobile robot to cross over obstacles along a path roughly determined either through teleoperation or by a navigation path-finder. Then, they present a simple but efficient control scheme derived from these concepts, taking into account ground clearance, vehicle safety, and possible occlusions in the vision field. This control scheme is applied to Rami, a four-tiltable-track robot equipped with force sensors, an inertial reference system, a laser-stripe range finder, and extensive real-time computing facilities based on a decentralized architecture.>

The motion planning problem is discussed for a mobile robot with a kinematic constraint, that is, the number of its degrees of freedom is less than the dimension of its configuration space. Addressing first the local problem of turning in a corridor, it is shown that sliding continuously on the outer wall in a turn constitutes a canonical contract trajectory, in the sense that if it is not safe, there cannot be another safe trajectory with no backing-up maneuver. For the case in which no smooth trajectory exists, some simple maneuvers are proposed. A heuristic path-planning algorithm based on a decomposition of free space into cones connected by turns is then presented. It is shown how this approach can be used within the more realistic framework of an actual mobile robot at work by implementing the results as a set of rules of behavior for the robot.>

This paper presents an alternative to the Potential Field Method of computing local collision-free motions for general manipulators. The main distinction is that we separate the description of the task from constraints of anti-collision. This enables to control accurately all relevant measures of the problem. A task is expressed by the minimization of a function plus eventually some geometric constraints, whereas anti-collision is translated into very simple linear constraints through the methods of the velocity dampers and the tangent separating planes. This approach is applied to the control of manipulators with a high number of degrees of freedom, using hierarchical descriptions of the environment and the robots. This is illustrated by two realizations, path planning for a ten link manipulator in the cluttered environment of a nuclear plant reactor, and cooperative tasks between two six degree of freedom robots. In the end we discuss how to incorporate the action of a global planner in this general framework.

We describe a robot system capable of locating a part in an unstructured pile of objects, choose a grasp on the part, plan a motion to reach the part safely, and plan a motion to place the part at a commanded position. The system requires as input a polyhedral world model including models of the part to be manipulated, the robot arm, and any other fixed objects in the environment. In addition, the system builds a depth map, using structured light, of the area where the part is to be found initially. Any other objects present in that area do not have to be modeled.

Regrasping must be performed whenever a robot's grasp of an object is not compatible with the task it must perform. This paper presents an approach to the problem of regrasping for a robot arm equipped with parallel-jaw end-effector. The method employs a table surface to place the object in intermediate positions.

We present a local method for obstacle avoidance based on the existence of extreme separating hyperplanes. Its application is then extended to the coordinated motion of several mobile robots, and to the avoidance of two manipulators. The techniques of local planning we develop here prove to be an interesting substitute to the potential field method, as they provide a simple geometric support for the analysis.