Low-cost fuzzy-based obstacle avoidance method for autonomous agents
Chapter in Scopus
- Additional Document Info
- View All
© Springer International Publishing Switzerland 2014. In this article, there lies an implementation of a novel fuzzy-based methodology that will reduce the amount of operations per iteration, reducing the computational cost of traditional fuzzy systems. This was achieved by implementing adaptive control theories like gain scheduling. Also, the method was implemented with a self-tuning technique using Genetic Algorithms (GA) that increases the method¿s efficiency, improving the following metrics: amount of collisions and time spent to finish the task. Two experiments were implemented to test the previously mentioned metrics; the first experiment will test the agent with several intersecting dynamic obstacles, in a long path. The second experiment will deal with two aggressive agents that will attempt to collide with the test agent. For comparison purposes, the Potential Fields¿ Method (PFM) was implemented and tested under the same metrics and experiments. The obtained results show for experiment 1 an improvement on the average collisions by 25.03% less compared to the PFM statistics and only and increase of 1.43% of the average time spent. And for experiment 2, there was an improvement of 93.46% of less average collisions compared to PFM and a reduction of 10.27% of average time spent. The contribution of this work is to implement a faster and less processing expensive method than traditional fuzzy ones.