EdgeNavMamba: Mamba Optimized Object Detection for Energy Efficient Edge Devices

2025-10-16

AI Summary

To address the challenge of balancing real-time performance and detection accuracy for resource-constrained edge devices in autonomous navigation, this paper proposes a lightweight Mamba-based object detection framework integrating knowledge distillation and reinforcement learning for end-to-end goal-directed navigation. Innovatively, we adapt the state-space model (Mamba) to edge vision perception tasks, designing an efficient student network trained on a custom indoor shape detection dataset. Detection outputs directly inform RL policies deployed in MiniWorld and IsaacLab. Experimental results demonstrate a 31% reduction in model parameters and 67% compression in model size; inference energy consumption drops by 73% on Jetson Orin Nano and Raspberry Pi 5. Navigation success rate exceeds 90% in MiniWorld, while detection accuracy maintains state-of-the-art (SOTA) performance.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Object Detection & Categorization
• Intelligent Robots: Localization, Mapping, and Navigation

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Deployment of efficient and accurate Deep Learning models has long been a challenge in autonomous navigation, particularly for real-time applications on resource-constrained edge devices. Edge devices are limited in computing power and memory, making model efficiency and compression essential. In this work, we propose EdgeNavMamba, a reinforcement learning-based framework for goal-directed navigation using an efficient Mamba object detection model. To train and evaluate the detector, we introduce a custom shape detection dataset collected in diverse indoor settings, reflecting visual cues common in real-world navigation. The object detector serves as a pre-processing module, extracting bounding boxes (BBOX) from visual input, which are then passed to an RL policy to control goal-oriented navigation. Experimental results show that the student model achieved a reduction of 67% in size, and up to 73% in energy per inference on edge devices of NVIDIA Jetson Orin Nano and Raspberry Pi 5, while keeping the same performance as the teacher model. EdgeNavMamba also maintains high detection accuracy in MiniWorld and IsaacLab simulators while reducing parameters by 31% compared to the baseline. In the MiniWorld simulator, the navigation policy achieves over 90% success across environments of varying complexity.

Problem

• Optimizing object detection for energy-efficient edge devices
• Reducing model size and energy consumption while maintaining accuracy
• Enabling real-time autonomous navigation on resource-constrained hardware

Innovation

• Mamba object detection model for efficient edge computing
• Reinforcement learning policy for goal-directed navigation control
• Custom dataset for training in diverse indoor environments