Returning home after an outward journey is a skill important to the survival of many insects and other animals. Many hymenopterans perform homing navigation using snapshot images taken at specific locations. Snapshot images can be processed in various ways, and this article focuses on the vector model of landmark distribution. In this article, we suggest a new landmark vector model with quantized distance information, and this method is highly successful for navigation without using a reference compass. The landmark distances based on egomotion can be obtained and this information produces a new type of homing navigation with landmark vectors. It includes varying sizes of landmark vectors, depending on the distance. We investigate the effect of quantized distance information or continuous distance in the landmark vectors for homing navigation performance. Even a few classes of discretization of landmark distances and a rough estimation of the current agent location can show excellent homing performance. Our experimental results are compared with those of other conventional navigation algorithms.
All Science Journal Classification (ASJC) codes
- Experimental and Cognitive Psychology
- Artificial Intelligence