NAR-*ICP: Neural Execution of Classical ICP-based Pointcloud Registration Algorithms
This work addresses the interpretability gap in neural networks for robotics applications, though it appears incremental as it extends existing benchmarks and methods.
The authors tackled the problem of combining neural networks with classical robotics algorithms by proposing NAR-*ICP, a GNN-based framework that learns intermediate computations of ICP-based pointcloud registration algorithms, achieving superior performance that surpasses the original algorithms on real-world and synthetic datasets.
This study explores the intersection of neural networks and classical robotics algorithms through the Neural Algorithmic Reasoning (NAR) blueprint, enabling the training of neural networks to reason like classical robotics algorithms by learning to execute them. Algorithms are integral to robotics and safety-critical applications due to their predictable and consistent performance through logical and mathematical principles. In contrast, while neural networks are highly adaptable, handling complex, high-dimensional data and generalising across tasks, they often lack interpretability and transparency in their internal computations. To bridge the two, we propose a novel Graph Neural Network (GNN)-based framework, NAR-*ICP, that learns the intermediate computations of classical ICP-based registration algorithms, extending the CLRS Benchmark. We evaluate our approach across real-world and synthetic datasets, demonstrating its flexibility in handling complex inputs, and its potential to be used within larger learning pipelines. Our method achieves superior performance compared to the baselines, even surpassing the algorithms it was trained on, further demonstrating its ability to generalise beyond the capabilities of traditional algorithms.