9 Conclusion

9 Conclusion

Implementing a mobile robot programming and control language requires a significant expenditure of effort. The hardware and software capabilities of the robot must be ascertained before any specifications can be created. Expected limitations must be expressed succinctly, but should not overshadow any positive claims. Difficulties must be surmounted as they occur and specifications might be changed to meet new requirements. When working in a team, members must be aware of accomplishments and duplicate effort must be avoided. As the system nears completion, refinement becomes necessary. Finally, users test the system and explain any errant behavior. Ideally, problems are fixed and the system is complete. Realistically, further improvements may be necessary for many different reasons-these remain for future robo-nauts, but the tools and information must be available. This thesis not only presents a new approach to mobile robot control, but suggests ways to improve LLAMA, if only indirectly. LLAMA meets the requirements of being a solid base for future work and daily results from its use validate this observation.

The FORTH-based interpreter is just one of a number of possible interpreter and compiler "front-ends" for LLAMA. The system was designed to accept new or improved interpreters. The sensor blackboard is not complete in that only four sensors have been placed on the blackboard. The blackboard could be improved and made more efficient if the "busy-wait" design was updated into a "sleep-wait" design. Local variables in each interpreter could be implemented if a temporary dictionary were used for each spawned interpreter. The dictionary would contain local variable and constant definitions. Arrays and floating point are not implemented in LLAMA, but could be implemented using FORTH techniques. Many other enhancements are possible.


Return to Thesis Title