Energy-Efficient Autonomous Navigation Benchmarking

A freshman computer engineering student at The Ohio State University is tackling a critical challenge in robotics: energy efficiency. Rocky Shao, with support from Dr. Marco Brocanelli's lab, is building an open-source research platform to systematically profile the energy-to-performance trade-offs of autonomous systems. The hardware setup is a modified RC car equipped with high-fidelity sensors—an Intel RealSense D455 for stereo depth and visual SLAM, and an RPLidar A2M8 for 2D obstacle detection—all powered by an NVIDIA Jetson Nano for onboard AI inference. The core innovation is the integration of real-time power monitoring to log consumption data as the system navigates.

Shao's framework runs on ROS 2 (Robot Operating System) and is designed to test how variables like CPU frequency scaling and edge computing offloading affect both performance and power draw. In his initial progress log, he detailed overcoming dependency issues with ROS 2 Jazzy and successfully publishing raw sensor data from the RealSense camera and LiDAR to ROS topics. The next phase involves attaching a power-measuring device to the battery and running autonomous navigation policies that fuse camera images, depth maps, IMU data, and LiDAR scans. The project aims to find the operational 'sweet spot' that maximizes navigation capability while minimizing energy use, a key concern for battery-powered drones, rovers, and mobile robots.