autonomous weapon turret

Conclusions

Performance Constraints/Bottlenecks

Several factors limited the performance of the AWT system during system testing. One limiting constraint was the low-budget webcam used to perform for the AWT’s image processing. Even at its max 30 fps rating, the webcam would not have been sufficient for a high-quality real-time application.

The image processing algorithm itself, however, turned out to be the biggest bottleneck in the AWT system. The task of processing large datasets in a real-time system is difficult and requires large amounts of memory as well as a high clock frequency. Ideally, the image processing algorithms should be parallelized, and the system should be run on a multi-core processer.

Future Work

There is a good deal of work that can be done to improve the performance of this system. The most obvious option for performance improvement would be to improve the quality of the hardware (webcam, computer, and motors). However, this will obviously add significant cost to the system.

Another path towards improvement is mentioned above in Performance Constraints/Bottlenecks. By parallelizing the image processing and using a multi-core system, significant improvements in performance will be seen.