인하대학교 첨단분야 혁신융합대학 미래자동차 사업단 - 첨단분야 혁신융합대학 미래자동차 사업단

Human-following autonomous mobile robots (wheelded, flying)

관리자 2024.03.19 15:35 244

Human-following autonomous mobile robots (wheelded, flying)

groupTeam Project Name
access_timeMeeting Schedule
Tue 19,20(18:00~18:50)
businessLocation
Yonghyeon Campus. 60-107
faceTeam Advisors
kwangki Kim
bookAcademic Majors of Interest
Electrical, Electronic, Info & Comm, Computer, Smart Mobility, AI, Mechanical, Aerospace
location_cityRelated Companies
Naver, HMC, LG, Samsung, Mando, Continental, etc. Katech, KITECH, ETRI, KIMM, KARI, etc.
peoplePartner(s) and Sponsor(s)

Goals

1. Production of 1/10 scale racing car (renovation of RC car) - The production of the completed hardware platform is now complete, and considering the number of participants, we are going to consider producing one more vehicle HW. 2. Learn and implement sensor data collection and processing, state estimation, route generation, and optimal control technologies using autonomous driving simulators (F1TENTH, Autoware, CARLA) and autonomous flight simulators (AirSim). 3. Develop an algorithm to recognize and follow objects using sensor fusion (distance/direction sensor (UWB), inertial sensor (IMU), and vision sensor (Stereo Camera)). 4. A robot that autonomously follows a human by determining or estimating the distance and direction between the anchor (human) and the tag (robot) using UWB signals and applying way-point tracking techniques based on this is installed in Turtlebot3 implement

Issues Involved or Addressed

For autonomous driving * Perception - Sensor Fusion * Planning (judgment) - based on real-time optimization * Control - Predictive control * Learning - Reinforcement Learning * Embedded System Design (Embedded System Design) - ARM series (Arduino, Mbed, etc.) * Software (ROS(2), Python, C++) Learn the theory and implement it through practice.
In addition to developing a modular autonomous robot algorithm based on SLAM, Path/Motion Planning, and Control modules, an integrated autonomous robot algorithm based on Pixel-to-Torque-based end-to-end learning will be carried out together and compare their performances (the pros and cons).