Abstract:Autonomous system navigation is a well-researched and evolving field. Recent advancements in improving robot navigation have sparked increased interest among researchers and practitioners, especially in the use of sensing data. However, this heightened focus has also raised significant privacy concerns, particularly for robots that rely on cameras and LiDAR for navigation. Our innovative concept of Radio Frequency (RF) map generation through ray-tracing (RT) within digital twin environments effectively addresses these concerns. In this paper, we propose DT-RaDaR, a robust privacy-preserving, deep reinforcement learning-based framework for robot navigation that leverages RF ray-tracing in both static and dynamic indoor scenarios as well as in smart cities. We introduce a streamlined framework for generating RF digital twins using open-source tools like Blender and NVIDIA's Sionna RT. This approach allows for high-fidelity replication of real-world environments and RF propagation models, optimized for service robot navigation. Several experimental validations and results demonstrate the feasibility of the proposed framework in indoor environments and smart cities, positioning our work as a significant advancement toward the practical implementation of robot navigation using ray-tracing-generated data.

What problem does this paper attempt to address?

The main problems that this paper attempts to solve are privacy protection and data efficiency in robot navigation. Specifically, traditional sensor - based robot navigation methods (such as using cameras and LiDAR) have improved the navigation capabilities of robots, but they also bring significant privacy risks and high computational resource requirements. These problems prompt researchers to explore new solutions. ### Core Problems of the Paper 1. **Privacy Protection**: Traditional navigation sensors (such as cameras and LiDAR) may inadvertently capture personal privacy information, such as people's facial features, locations or behaviors, when collecting environmental data. This raises concerns about data security, user consent and potential misuse. 2. **Data Efficiency**: High - resolution sensors (such as RGB - D cameras and 3D LiDAR), although providing detailed environmental information, their high costs and large - scale data processing requirements limit their application scope. ### Solutions To solve the above - mentioned problems, the paper proposes a new framework named DT - RaDaR. This framework utilizes ray - tracing (RT) to generate radio - frequency (RF) digital twin environments and combines deep reinforcement learning (RL) to achieve privacy - protected robot navigation. Specific contributions are as follows: - **C1. High - Fidelity Digital Twin Creation**: Through the open - source tool Blender and its OpenStreetMap (OSM) plugin, create digital twin models that accurately simulate indoor and outdoor environments in the real world. - **C2. Precise Configuration of Scenes and Generation of Propagation Characteristics**: Use NVIDIA's Sionna RT software to apply differential ray - tracing (DRT) in the digital twin of each scene to generate propagation data. - **C3. Reinforcement Learning Algorithm Design**: Design an RL - based algorithm, use the propagation data generated from ray - tracing tools for custom environmental modeling, state extraction, action selection and reward function design, and train DQN agents to generate optimal navigation paths. - **C4. The First Ray - Tracing Digital Twin Dataset**: Construct a ray - tracing digital twin dataset covering indoor laboratory environments and outdoor smart city scenes, which specifically simulate downtown Dallas and downtown Houston. - **C5. Framework Verification**: Through strict experimental verification, the effectiveness of the DT - RaDaR framework is proved. Experimental results show that the time required for ray - tracing - based training is reduced by approximately 74.4% compared with the method using LiDAR/camera sensors. ### Summary The DT - RaDaR framework aims to effectively solve the problems of privacy protection and data efficiency in robot navigation by introducing ray - tracing technology and digital twin models. This framework not only improves the safety and efficiency of navigation but also provides new ideas and directions for the future development of robotics.

DT-RaDaR: Digital Twin Assisted Robot Navigation using Differential Ray-Tracing

Real-Time Navigation In Dynamic Human Environments Using Optimal Reciprocal Collision Avoidance

Using Low-Power Radios for Mobile Robots Navigation

AI-based UAV navigation framework with digital twin technology for mobile target visitation

RNR-Nav: A Real-World Visual Navigation System Using Renderable Neural Radiance Maps

iNAV-drlSLAM: An Improved Indoor Self-Driving Framework for Mobile Robots using Deep Reinforcement Learning integrated with SLAM

Real-Time Navigation for Bipedal Robots in Dynamic Environments

Online Robot Navigation and Manipulation with Distilled Vision-Language Models

Enhancing Autonomous Navigation by Imaging Hidden Objects using Single-Photon LiDAR

Beyond Uncertainty: Risk-Aware Active View Acquisition for Safe Robot Navigation and 3D Scene Understanding with FisherRF

NavDreams: Towards Camera-Only RL Navigation Among Humans

A Novel Artificial Organic Control System for Mobile Robot Navigation in Assisted Living Using Vision- and Neural-Based Strategies

Autonomous Navigation System of Indoor Mobile Robots Using 2D Lidar

High-Speed Robot Navigation using Predicted Occupancy Maps

PANav: Toward Privacy-Aware Robot Navigation via Vision-Language Models

Visibility-Aware RRT* for Safety-Critical Navigation of Perception-Limited Robots in Unknown Environments

Autonomous Navigation by Mobile Robot with Sensor Fusion Based on Deep Reinforcement Learning

Real-Time Object Navigation With Deep Neural Networks and Hierarchical Reinforcement Learning

Ad Hoc-Obstacle Avoidance-Based Navigation System Using Deep Reinforcement Learning for Self-Driving Vehicles

CoverNav: Cover Following Navigation Planning in Unstructured Outdoor Environment with Deep Reinforcement Learning

RoadRunner -- Learning Traversability Estimation for Autonomous Off-road Driving