좁은 통로와 장애물로 구성된 환경에서의 군집 지상-공중 이중 모드 로봇 경로계획 및 제어 = Path Planning and Control for Swarm of Ground-aerial Dual-mode Robots in Environments with Narrow Passages and Obstacles

다국어 초록 (Multilingual Abstract)

Ground–aerial dual-mode robots have been extensively studied because they can overcome the operational limitations of single-mode robots by simultaneously achieving energy efficiency and maneuverability. With a quadcopter equipped with wheels, such robots can conserve energy through ground locomotion on flat terrain and switch to aerial mode for obstacle avoidance or high-speed transit, enabling efficient operation. When extended to multi-robot systems, these dual-mode platforms can operate more efficiently than swarms composed of single-mode robots, as they can adaptively switch modes according to situational requirements. However, prior research on ground–aerial dual-mode robots that utilize only propeller thrust for both ground and aerial locomotion has largely been limited to single-robot scenarios or to multi-robot systems in which ground propulsion is provided by dedicated wheel motors.
Therefore, this study proposes path planning and control methods for a swarm of ground–aerial dual-mode robots that perform both ground and aerial locomotion using propeller thrust alone, taking into account robot-specific characteristics in environments with narrow corridors and ground obstacles. To this end, the angular velocities of the robot rotors are measured in simulation during ground locomotion, aerial flight, and ground waiting, and the unit energy consumption for each locomotion mode is calculated and incorporated into the cost function used for path planning. In addition, an aerial locomotion weight parameter is introduced into the cost function to enable the generation of paths that prioritize either energy efficiency or mission time reduction. For collision-free path planning among multiple robots, a priority-based sequential planning scheme is employed, where robots plan their paths in the order of predefined priorities and lower-priority robots treat the paths of higher-priority robots as dynamic obstacles. The generated paths are refined through a post-processing step, and once all paths have been obtained, they are tracked using Model Predictive Control (MPC), a look-up table containing experimentally derived thrust and pitch angle data, and low-level PID controllers.
To evaluate the proposed system, simulation experiments were conducted across multiple scenarios. In all scenarios, it was confirmed that the planner can generate paths that emphasize either energy efficiency or time reduction depending on the aerial locomotion weight setting, and that all robots safely reached their goals while avoiding inter-robot collisions from start to goal.

목차 (Table of Contents)

• 제1장 서론 1
• 제1절 연구 배경 1
• 제2절 관련 연구 4
• 제3절 연구 목표 및 연구 방법 8
• 제2장 이론적 배경 9
• 제1절 Voxel Map & Octomap 9
• 제2절 A* 경로계획 알고리즘 10
• 제3절 모델 예측 제어 (Model Predictive Control, MPC) 11
• 제3장 제안하는 방법 12
• 제1절 문제 정의 12
• 제2절 군집 지상-공중 이중 모드 로봇 경로계획 12
• 3.2.1 제안하는 경로계획 방법의 전체 구조 12
• 3.2.2 에너지 소모량을 고려한 초기 전역 경로 생성 13
• 3.2.2 초기 전역 경로 후처리 18
• 제3절 지상-공중 이중 모드 로봇 제어 시스템 19
• 3.3.1 모드별 제어 시스템 설계 19
• 제4장 실험 환경 구축 및 결과 21
• 제1절 실험 환경 21
• 4.1.1 실험 환경 구축 및 평가 지표 설정 21
• 4.1.2 제어 시스템 성능 테스트 22
• 4.1.3 실험 시나리오 구성 24
• 제2절 실험 결과 26
• 4.2.1 시나리오 1번 실험 결과 26
• 4.2.2 시나리오 2번 실험 결과 31
• 제3절 지상-공중 이중 모드 로봇 하드웨어 36
• 4.2.3 기체 제작 36
• 4.2.4 호버링 테스트 38
• 제5장 결론 40
• 참고문헌 41
• Abstract 44