NanoSLAM: Enabling Fully Onboard SLAM for Tiny Robots

2023-09-21

IEEE Internet of Things Journal

AI Summary

To address the challenge of implementing fully onboard simultaneous localization and mapping (SLAM) on centimeter-scale microrobots constrained by severe computational and memory resources, this paper proposes a lightweight end-to-end SLAM architecture tailored for the ultra-low-power RISC-V parallel processor GAP9. The architecture integrates parallelized feature extraction, a memory-aware sparse bundle adjustment (BA) solver, and compute–memory co-optimized compression techniques. It achieves full-pipeline SLAM execution at only 87.9 mW power consumption. Evaluated on a 44 g nano-drone, the system attains mapping accuracy of 4.5 cm and end-to-end latency under 250 ms. This work marks the first demonstration of milliwatt-level, fully onboard, real-time SLAM—establishing a deployable localization and mapping paradigm for micro-autonomous robots.

Abstract

Perceiving and mapping the surroundings are essential for autonomous navigation in any robotic platform. The algorithm class that enables accurate mapping while correcting the odometry errors present in most robotics systems is simultaneous localization and mapping (SLAM). Today, fully onboard mapping is only achievable on robotic platforms that can host high-wattage processors, mainly due to the significant computational load and memory demands required for executing SLAM algorithms. For this reason, pocket-size hardware-constrained robots offload the execution of SLAM to external infrastructures. To address the challenge of enabling SLAM algorithms on resource-constrained processors, this article proposes NanoSLAM, a lightweight and optimized end-to-end SLAM approach specifically designed to operate on centimeter-size robots at a power budget of only 87.9 mW. We demonstrate the mapping capabilities in real-world scenarios and deploy NanoSLAM on a nano-drone weighing 44 g and equipped with a novel commercial RISC-V low-power parallel processor called GAP9. The algorithm, designed to leverage the parallel capabilities of the RISC-V processing cores, enables mapping of a general environment with an accuracy of 4.5 cm and an end-to-end execution time of less than 250 ms.

Problem

• Enabling SLAM on tiny robots with limited power
• Reducing computational load for onboard SLAM execution
• Achieving accurate mapping on resource-constrained processors

Innovation

• Lightweight end-to-end SLAM for tiny robots
• Optimized for RISC-V low-power parallel processors
• Enables onboard mapping with 87.9 mW power