Semester

Summer

Date of Graduation

2024

Document Type

Thesis

Degree Type

MS

College

Statler College of Engineering and Mineral Resources

Department

Mechanical and Aerospace Engineering

Committee Chair

Yu Gu

Committee Co-Chair

Jason Gross

Committee Member

Jason Gross

Committee Member

Guilherme Pereira

Abstract

The advancement of autonomous search holds significant promise for applications ranging from emergency response to planetary exploration. This thesis investigates strategies to enhance autonomous search performance in large-scale environments. The main contribution of this work is its practical application in real-world scenarios, where efficient search methods are essential for managing vast amounts of data, particularly in large environments. Effective search planning requires navigating complexities such as limited prior information and managing large state spaces, necessitating advanced strategies to plan with this limited information. Additionally, balancing exploration and exploitation is crucial for optimizing the search process, as it ensures thorough coverage of the search area while efficiently utilizing resources. These challenges highlight the need for innovative approaches as traditional methods often fall short in extensive environments. State-of-the-art tools such as Monte Carlo Tree Search (MCTS) play a crucial role in advancing autonomous search technologies. MCTS is effective at handling large, complex state spaces and balancing exploration and exploitation, making it well-suited for scenarios with vast search spaces and limited prior information. However, MCTS can be computationally expensive and limited in its planning horizon, leading to potential suboptimal decisions in long-term scenarios. To address these challenges, this thesis introduces a novel approach that formulates the search problem as a belief Markov decision process with options (BMDP-O). This formulation integrates sequences of actions to navigate between regions of interest (ROIs), enabling efficient scaling to large environments. This allows for more efficient planning by breaking down the search into manageable segments and focusing resources on the most promising regions. The proposed method proves particularly efficient in environments with multiple ROIs. A robot can search within an ROI and transition to nearby ROIs without extensive searching in low-probability areas. This strategy ensures a balanced and efficient search in extensive environments. Results demonstrate that this method finds objects faster in large environments compared to traditional MCTS while still maintaining computational efficiency.

Included in

Robotics Commons

Share

COinS