SCV - Smart Companion Vehicle

자연어 명령 기반 이동형 AI 홈 어시스턴트

2026.05TEAM A207Autonomous Driving, ROS2 Integration, LLM Drive Bridge, Runtime Tools

프로젝트 개요

자연어 명령을 실행 가능한 주행·정렬·조작 흐름으로 변환하고, ROS2/Nav2 주행과 비전 정렬, 로봇팔 런타임을 연결한 Smart Companion Vehicle 프로젝트입니다.

주요 성과

  • RoboCrew System2 planner → System1 packet server → ROS2 drive bridge로 자연어 명령 실행 경로 구성
  • Nav2/AMCL/SLAM 기반 실내 주행과 motion_coordinator, trolley_motion_adapter 제어 경로 통합
  • Yahboom mecanum base USB serial runtime, BNO085 IMU, RPLiDAR/RealSense bringup 정리
  • vision alignment, AprilTag 기반 정밀 정렬, 로봇팔 runtime helper를 서비스 시나리오에 연결
  • stub-hardware smoke mode, web teleop, map helper로 반복 검증 가능한 시연 운영 환경 구성

실행 파이프라인

App / Voice사용자 명령
RoboCrew System2LLM planner
System1 Packet Server실행 packet
ROS2 Drive BridgeNav2 / relative motion
Motion Adaptercmd_vel / motion mode
Yahboom Runtimemecanum base

핵심 패키지

패키지역할
trolley_driverYahboom UART driver, mecanum IK, encoder odom
trolley_nav_bringupNav2, SLAM, AMCL, sensor launch
robocrew_ros_adapterRoboCrew packet drive bridge
motion_coordinatormotion mode FSM과 cmd_vel 중재
trolley_motion_adapterLLM drive stack command adapter

본인 담당 업무

  • Autonomous Driving 역할 수행
  • ROS2 Integration 역할 수행
  • LLM Drive Bridge 역할 수행
  • Runtime Tools 역할 수행

트러블슈팅

  • CAN/F446ZE gateway는 별도 feature 단계로 분리하고, 실제 runtime 경로는 USB 직결 Yahboom base 기준으로 정리
  • 실기 반복 튜닝의 비용을 줄이기 위해 stub-hardware smoke mode와 web teleop/map helper를 함께 구성

추가 개발 예정

  • auto initial pose
  • CAN/F446ZE gateway production integration
  • precision calibration 고도화

갤러리