Remote-control versus autonomy

Assuming that the robot may roam over a larger area than the matched zone, a locomotion method is needed. This implies that the user controls the robot motion through an interface. In Section 10.2, locomotion was presented for navigating in a large virtual world and was explained as controlling a cart (Figure 10.5). The robot in the real world behaves geometrically like the cart in the pure virtual world; however, some differences are: 1) The robot cannot simply teleport to another location. It is, however, possible to connect to a different robot, if many are available, which would feel like teleportation to the user. 2) The robot is subject to constraints based on its physical design and its environment. It may have rolling wheels or walking legs, and may or may not be able to easily traverse parts of the environment. It will also have limited driving speed, turning speed, and battery life. 3) A high cost is usually associated with crashing the robot into people or obstacles.

A spectrum of choices exists for the user who teleoperates the robot. At one extreme, the user may continuously control the movements, in the way that a radio-controlled car is driven using the remote. Latency becomes critical some applications, especially telesurgery [190,362]. At the other extreme, the user may simply point out the location on a map or use a virtual laser pointer (Section 10.2) to point to a visible location. In this case, the robot could execute all of the motions by itself and take the user along for the ride. This requires a higher degree of autonomy for the robot because it must plan its own route that accomplishes the goals without running into obstacles; this is known in robotics as motion planning [165]. This frees the user of having to focus attention on the minor robot movements, but it may be difficult to obtain reliable performance for some combinations of robot platforms and environments.

Steven M LaValle 2020-01-06