Development of a Situational Awareness Graph for Enhanced Robotic Search and Rescue Operations
This project involves the development and implementation of an advanced situational awareness system for a robotic platform, specifically designed for autonomous exploration, mapping, and classification within search and rescue operations. The thesis will be developed in collaboration with the Robotics group of the Royal Dutch Marechaussee, where the student will regularly spend part of the project time. The core objective is to create a dynamic situational graph of the environment, enabling the robot to navigate complex terrains and locate individuals, classifying them based on their risk profile, such as wounded or active.
Preliminary work has been conducted using the Spot robot from Boston Dynamics, equipped with a 360-degree camera, LIDAR, GPS, and running on the Robot Operating System (ROS). Autonomous exploration capabilities have been established, forming a solid foundation for further development.
Project Goals:
- Situational Awareness Graph Development: Enhance the robot's understanding of its environment by constructing a situational awareness graph that dynamically updates as the robot explores. This graph should detail terrain features, obstacles, and the location of persons within the area.
- Person Detection and Classification: Implement algorithms capable of detecting individuals within the robot's environment and classifying them into categories based on their observed risk profile (e.g., wounded, incapacitated, or active), utilizing data from the robot's sensors.
- Integration and Testing: Integrate the developed system with the existing autonomous exploration capabilities of the Spot robot, ensuring seamless operation and real-time data processing.
- Evaluation: Conduct comprehensive testing to evaluate the system's effectiveness in accurately mapping environments, detecting and classifying individuals, and navigating autonomously in simulated search and rescue scenarios.
Requirements:
- Technical Skills: applicants must demonstrate a strong proficiency in Python, with knowledge of C++ being highly advantageous. A solid understanding and hands-on experience with ROS are essential for this project.
- Analytical Skills: ability to analyze complex data from multiple sensors and develop algorithms for real-time processing and decision-making.
- Location and Commitment: this project will be based in The Hague, with a minimum on-site commitment of 2 days per week to facilitate direct interaction with the research team and access to the robotic platform.
Similar work:
- Kimera: https://arxiv.org/abs/1910.02490
- S-Graphs+: https://arxiv.org/abs/2212.11770
- Khronos: https://arxiv.org/abs/2402.13817
- Situational graphs: https://arxiv.org/abs/2202.12197