Multi-agent (MA) AI holds great promise for enhancing edge devices with limited computing resources [1]. In particular, MA simultaneous localization and mapping (SLAM) is actively under investigation to improve map accuracy in swarm robotics. Conventional keyframe-based SLAM, leveraging landmarks [2]–[4], provides the appropriate map accuracy but is unsuitable for MA SLAM on decentralized edge devices due to computational complexity. The neuromorphic SLAM is a candidate for MA SLAM owing to its low complexity in singleagent operation [5]. However, this method is still infeasible in MA SLAM due to the drastic increase of complexity in MA map correction. As such, several challenges need to be addressed via circuit-algorithm co-design in deploying MA SLAM to edge devices. In this paper, we present the BEE-SLAM accelerator, inspired by bee communication, featuring hybrid mixed-signal/digital biomimetic circuits and MA map error correction (MAEC), achieving the energy efficiency of 17.96 TOPS/W in outdoor MA SLAM operation.
No takes yet. Share an insight, caveat, or question.
Lee et al. (2024) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: