The Robotic cylindrical maze, developed in an experiment led by Alejandra Barrera et al. (2011), aimed to compare spatial cognition between rats and robots through goal-oriented tasks.

In training, robots utilized Hebbian and reinforcement learning, specifically Actor-Critic architecture, to learn the optimal path from a fixed starting point to the goal within the maze.

During testing, robots applied their cognitive map, previously developed with expectations of maximum reward, to successfully navigate from various starting positions to the goal.

Explore Mazeengineers’ robotic cylindrical maze, adaptable for both mice and rats, with options for custom coloring and customization upon request.