A Novel ViDAR Device With Visual Inertial Encoder Odometry and Reinforcement Learning-Based Active SLAM Method

2025-06-16

IEEE Transactions on Industrial Informatics

AI Summary

To address the challenges of encoder integration difficulty, weak active perception capability, and high cost/complexity in multi-sensor SLAM, this paper proposes the ViDAR hardware architecture—comprising a monocular camera, IMU, and motor encoder—along with a tightly coupled Visual-Inertial-Encoder Odometry (VIEO) model. Furthermore, we design a Platform-Motion-Decoupled Deep Reinforcement Learning (PPO-based) active SLAM framework, wherein state representation decoupling separates motion policy learning from perceptual modeling. Key innovations include: (i) the first joint calibration and modeling method for ViDAR hardware and VIEO; and (ii) a motion-decoupled DRL active strategy to enhance feature diversity and inter-frame co-visibility. Experiments demonstrate that VIEO significantly improves both the number of co-visible features and state estimation accuracy over conventional VIO. The DRL policy increases feature diversity weight by 37% and reduces absolute pose error by 28%.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Multi-modal Vision
• Machine Learning: Multi-instance/Multi-view Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

In the field of multi-sensor fusion for simultaneous localization and mapping (SLAM), monocular cameras and IMUs are widely used to build simple and effective visual-inertial systems. However, limited research has explored the integration of motor-encoder devices to enhance SLAM performance. By incorporating such devices, it is possible to significantly improve active capability and field of view (FOV) with minimal additional cost and structural complexity. This paper proposes a novel visual-inertial-encoder tightly coupled odometry (VIEO) based on a ViDAR (Video Detection and Ranging) device. A ViDAR calibration method is introduced to ensure accurate initialization for VIEO. In addition, a platform motion decoupled active SLAM method based on deep reinforcement learning (DRL) is proposed. Experimental data demonstrate that the proposed ViDAR and the VIEO algorithm significantly increase cross-frame co-visibility relationships compared to its corresponding visual-inertial odometry (VIO) algorithm, improving state estimation accuracy. Additionally, the DRL-based active SLAM algorithm, with the ability to decouple from platform motion, can increase the diversity weight of the feature points and further enhance the VIEO algorithm's performance. The proposed methodology sheds fresh insights into both the updated platform design and decoupled approach of active SLAM systems in complex environments.

Problem

• Enhance SLAM performance with motor-encoder integration
• Improve active capability and FOV cost-effectively
• Develop decoupled active SLAM using reinforcement learning

Innovation

• ViDAR device with visual-inertial-encoder odometry
• DRL-based active SLAM decoupling platform motion
• Calibration method for accurate VIEO initialization