日本フィジカルAI新聞

世界のフィジカルAIを、日本語で。

自律移動arXiv:2002.11913

障害物検知・回避機能を備えた最短経路探索型自動配送ロボットAlpha-N

Alpha-N: Shortest Path Finder Automated Delivery Robot with Obstacle Detection and Avoiding System

シェア:XThreadsFacebookLINEはてブBluesky

ダイクストラ法とRFIDタグで最短経路を計画し、Faster R-CNNで障害物を検知・回避しながら自律走行する配送ロボットを開発・評価した論文。

著者: A. A. Neloy, R. A. Bindu, S. Alam, R. Haque, M. Saif, A. Khan, N. M. Mishu, S. Siddique

分類: cs.RO

原文アブストラクト

Alpha N A self-powered, wheel driven Automated Delivery Robot is presented in this paper. The ADR is capable of navigating autonomously by detecting and avoiding objects or obstacles in its path. It uses a vector map of the path and calculates the shortest path by Grid Count Method of Dijkstra Algorithm. Landmark determination with Radio Frequency Identification tags are placed in the path for identification and verification of source and destination, and also for the recalibration of the current position. On the other hand, an Object Detection Module is built by Faster RCNN with VGGNet16 architecture for supporting path planning by detecting and recognizing obstacles. The Path Planning System is combined with the output of the GCM, the RFID Reading System and also by the binary results of ODM. This PPS requires a minimum speed of 200 RPM and 75 seconds duration for the robot to successfully relocate its position by reading an RFID tag. In the result analysis phase, the ODM exhibits an accuracy of 83.75 percent, RRS shows 92.3 percent accuracy and the PPS maintains an accuracy of 85.3 percent. Stacking all these 3 modules, the ADR is built, tested and validated which shows significant improvement in terms of performance and usability comparing with other service robots.

関連論文

  1. 1Boston Dynamics、車輪付きAtlasの研究試作機を紹介研究・技術
  2. 2newmoとKDDIが資本業務提携 2028年に堺市で自動運転タクシー商用化目指す提携・買収
  3. 3GXOとApptronik、Apolloの物流現場試験を拡大 最大1,000台の展開を構想量産・展開
  4. 4NVIDIA、CoRLでGR00T 2とIsaac Lab 3.0を発表新製品・発表
  5. 5UC BerkeleyとNVIDIAの研究者、AIエージェントでロボット制御コードを仮想環境で改善研究・技術

PR・アフィリエイト広告

MINISFORUM(ミニPC)

MINISFORUM(ミニPC)の公式広告バナーMINISFORUM公式ストアを見る ↗
PR本紙発行元 EmplifAI