40.8 CONCLUSIONS

---

This is the most interesting section, but unfortunately, very vague. In the next few years Powerful computer hardware advances will allow low cost, but extremely powerful operations. This will make many memory intensive path planning methods feasible, the very complex operations will now be reasonable for calculation. If computers become powerful enough, then an expert path planning system would be a definite asset, making the robots flexible over a number of different frontiers, without any programming. The advance of customized computer hardware which now does graphics indicates that it is very feasible to convert complex robotic path planning from a software to a hardware domain. The potential of parallel processing also opens doors to solving very complex and involved problems.

A good direction for robotics research is to allow a path planning system which will combine the various problems and solutions. This will allow the robot to switch modes with an expert system and thus chose the solution to fit the problem, and not try to make the problem fit the solution, as is commonly done now. This would involve the use of rules and heuristics (based on the elements discussed in this paper) to build up a very complex and complete path planner from very simple methods (which are currently being researched). The multi-level path planning strategies are seen to also hold promise for robotic development (the author prefers the Dynamics Path Planning approach). These approaches allow the best of all planning methods.