Path-Planning and Collision Avoidance: Neural Network Approach
Studied learning-based path planning by reproducing Neural A* and U-Net–based planners and analyzing encoder–decoder variants.
Benchmark Reproduced key published results and benchmarked the architectures in PyTorch; proposed modifications for dynamic and multi-agent navigation.
- Reproduced key results from Neural A* and U-Net–based path planning studies, implementing and benchmarking architectures in PyTorch.
- Analyzed encoder–decoder variants (VGG-16, ResNet-50) and proposed modifications for dynamic and multi-agent navigation.
Skills