This project aims to develop an advanced walking simulator leveraging various artificial intelligence algorithms alongside the Pymunk physics engine. It creates a complex physical simulation environment where different animal models, defined by skeletal and muscular structures, can be optimized for enhanced movement across various terrains and obstacles.
- Physical Simulation Environment: Uses Pymunk to accurately simulate the physical laws affecting the movement of animal models.
- Walking Optimization: Implements multiple AI algorithms to optimize walking based on the distance covered and time.
- Visualization and Analysis: Ability to visualize walking movements and analyze the performance of each model.
This project includes several implementations of AI algorithms, each designed to optimize the walking process differently. These algorithms are organized into separate folders for easy navigation and experimentation:
- Q Learning: A reinforcement learning approach to adjust walking strategies.
- NEAT (NeuroEvolution of Augmenting Topologies): Evolves neural networks to improve movement coordination.
- Simulated Annealing: An optimization heuristic for finding the best configuration of walking parameters.
- Genetic Algorithm: Optimizes animal model characteristics through natural selection, mutation, and crossover.
In this simulation, I have incorporated two distinct animal structures to explore and optimize walking strategies under varying scenarios. Each structure is designed to mimic a different set of biological and mechanical properties, allowing me to test and refine the algorithms across a broader spectrum of walking dynamics. This approach enables me to simulate and analyze how different anatomical configurations impact the efficiency and adaptability of movement, providing valuable insights into the optimization processes of the AI algorithms.