next up previous contents
Next: 9.1 Remote Tele-Assistance Up: On-line Estimation of Visual-Motor Previous: 8.4 Visual Space Task

9 Future Work

The purpose of the methods we have developed is to make robotic manipulation possible outside of carefully engineered manufacturing applications. Instead we seek real world applications in unstructured environments. The experiments in this thesis have been done in a lab environment using industrial robot arms. In this kind of setting there is no particular point in using uncalibrated methods like the ones we have presented in this thesis. The robots are easy to calibrate (in fact they come with a calibration in the manufacturer supplied software), they are not mobile, and the environment around them is easy to engineer. This environment is not where we think our methods will find useful applications (but to us it was the only available testbed). Consider instead manipulation in human everyday environments. The manipulator is likely to have to be mobile. Unlike the case in a factory, objects in the outside world don't come on conveyor belts. A mobile manipulator on a mobile platform can't be as heavy and rigid as a fixed one, and a non-rigid manipulator is not easy to calibrate. The mobile manipulator also presents a real uncalibrated situation. The manipulator base is not fixed with respect to the world, so its world calibration changes all the time. And since the mobile manipulator can operate in a very large space, even if the environment is known and modelable, it may just be too large to make robotic manipulation worthwhile if a model has to be manually typed in.

We propose two real-world applications where we think that the integrated model estimation, control and visual simulation system described in this thesis is particularly suitable: Remote tele-assistance (intelligent tele-operation) of a mobile robot arm and hand, and a wheelchair mounted robot manipulator (external prosthesis) for aiding the handicapped in their daily chores. Next we will point out some interesting future research problems that we think are important for these applications.




next up previous contents
Next: 9.1 Remote Tele-Assistance Up: On-line Estimation of Visual-Motor Previous: 8.4 Visual Space Task

Martin Jägersand