Legged Robot State Estimation With Invariant Extended Kalman Filter Using Neural Measurement Network

TL;DR

This work develops a state estimation framework that trains a neural measurement network (NMN) to estimate the base's linear velocity and foot contact probability, which are then employed as measurements in an invariant extended Kalman filter.

Abstract

This paper introduces a novel proprioceptive state estimator for legged robots that combines model-based filters and deep neural networks. Recent studies have shown that neural networks such as multi-layer perceptron or recurrent neural networks can estimate the robot states, including contact probability and linear velocity. Inspired by this, we develop a state estimation framework that integrates a neural measurement network (NMN) with an invariant extended Kalman filter. We show that our framework improves estimation performance in various terrains. Existing studies that combine model-based filters and learning-based approaches typically use real-world data. However, our approach relies solely on simulation data, as it allows us to easily obtain extensive data. This difference leads to a gap between the learning and the inference domain, commonly referred to as a sim-to-real gap. We address this challenge by adapting existing learning techniques and regularization. To validate our proposed method, we conduct experiments using a quadruped robot on four types of terrain: \textit{flat}, \textit{debris}, \textit{soft}, and \textit{slippery}. We observe that our approach significantly reduces position drift compared to the existing model-based state estimator.

Figures (6)

• Figure 1: By combining our Neural Measurement Network (NMN) with a model-based filter, our method can reduce the position estimation error for the entire trajectory of the Raibo robot on an air mattress using only proprioceptive sensors by one-third compared to the purely model-based methods.
• Figure 2: Schematic of the proposed framework: To devise a learning-based state estimator, we first develop the policy for trajectory sampling. The NMN takes as input the proprioceptive sensor value and previous joint target and outputs contact probability and body linear velocity. These measurements are passed to the LPF and used to update the IEKF.
• Figure 3: Position estimation comparison: Proposed NMN vs. model-based GRF contact on four different terrains. Our NMN reduces the z-direction position drift compared to the model-based approach. Either method employs the slip rejection method.
• Figure 4: Experiment Terrain Overview: Upper left - Flat, Upper middle - Soft, Upper right - Slippery, Lower - Debris.
• Figure 5: We compare the body linear velocity error on the flat experiment for each ablation study.