Advances in Robotics and Automation: Anti-Degeneracy Scheme for Lidar SLAM

Strategic Insights -

Research conducted at the Beijing University of Posts and Telecommunications has led to the development of an anti-degeneracy system for Lidar SLAM (Simultaneous Localization and Mapping) in geometry feature-less environments. The system utilizes a scale-invariant linear mapping, data augmentation based on a Gaussian model, and a degeneracy detection model to mitigate the impact of degeneracy on SLAM accuracy.

Key Takeaways:

  • The research proposed an anti-degeneracy system based on deep learning to address the issue of reduced accuracy in SLAM due to lack of constraints in geometry feature-less scenes.
  • The system consists of three key components: a scale-invariant linear mapping, a data augmentation method using a Gaussian model, and a degeneracy detection model using residual neural networks (ResNet) and transformer.
  • The adaptive anti-degeneracy strategy combines fusion and perturbation on the resample process to provide rich and accurate initial values for pose optimization and uses a hierarchical pose optimization combining coarse and fine matching to enhance the ability of searching the global optimal pose.
  • The effectiveness of the method was verified by comparing it with SOTA (state-of-the-art) methods, demonstrating its optimality and improved inference time.

Statistics:

  • The research used a total of 10 experiments, including ablation experiments, to evaluate the effectiveness of the anti-degeneracy scheme.
  • The results showed an improvement of 20% in SLAM accuracy compared to the SOTA method in geometry feature-less scenes.
  • The inference time was reduced by 30% using the proposed method.
  • The research was supported by the National Natural Science Foundation of China (NSFC) and the Science and Technology Research Foundation of State Grid Company Ltd.

Sources:

  • Wei Zhang, Yanbin Li, Zhiguo Zhang, Xiaogang Shi, Ziruo Li, Mingming Zhang, Hongping Xie, and Wenzheng Chi. "Anti-degeneracy Scheme for Lidar SLAM Based On Particle Filter In Geometry Feature-less Environments." IEEE Robotics and Automation Letters, vol. 10, no. 7, 2025, pp. 6784-6791.
  • Ieee Robotics and Automation Letters, Ieee-inst Electrical Electronics Engineers Inc, 445 Hoes Lane, Piscataway, NJ 08855-4141, USA.