Semi-Autonomous Manipulation Pipeline (AIST, Japan)

CALL-M robot performing autonomous shelf-picking in convenience store environment

Research & Development | CNRS-AIST Joint Robotics Laboratory

Project Summary

Designed and developed an advanced semi-autonomous manipulation pipeline specifically for retail applications.

Technical Architecture

Core Technologies:

  • Point Cloud Library (PCL) - Advanced 3D perception and object segmentation

  • ROS2 Framework - Distributed robotics middleware with custom node orchestration

  • MoveIt2 - Sophisticated motion planning and collision avoidance

  • Docker Containers - Reproducible deployment and environment isolation

Perception Pipeline:

  • Point cloud processing and object segmentation

  • Pose estimation and grasp point calculation

  • Octomap plugin for obstacle detection and avoidance

Manipulation Strategy:

  • Adaptive grasp planning based on object geometry

  • Trajectory generation

  • Error recovery and re-planning capabilities

Research Contributions

Academic Publication “Development of a semi-autonomous manipulation pipeline for robotic shelf-picking operations”

Status: Under review for IEEE/SICE International Symposium on System Integration (SII2026)

Role: First Author