Author ORCID Identifier

https://orcid.org/0000-0001-8048-0523

Semester

Fall

Date of Graduation

2023

Document Type

Dissertation

Degree Type

PhD

College

Statler College of Engineering and Mineral Resources

Department

Mechanical and Aerospace Engineering

Committee Chair

Guilherme A S Pereira

Committee Member

Jason Gross

Committee Member

Yu Gu

Committee Member

Xi Yu

Committee Member

Adam Halasz

Abstract

This dissertation advances the field of autonomous vehicle motion planning in various challenging environments, ranging from flows and planetary atmospheres to cluttered real-world scenarios. By addressing the challenge of navigating environmental flows, this work introduces the Flow-Aware Fast Marching Tree algorithm (FlowFMT*). This algorithm optimizes motion planning for unmanned vehicles, such as UAVs and AUVs, navigating in tridimensional static flows. By considering reachability constraints caused by vehicle and flow dynamics, flow-aware neighborhood sets are found and used to reduce the number of calls to the cost function. The method computes feasible and optimal trajectories from start to goal in challenging environments that may contain obstacles or prohibited regions (e.g., no-fly zones). The method is extended to generate a vector field-based policy that optimally guides the vehicle to a given goal. Numerical comparisons with state-of-the-art control solvers demonstrate the method's simplicity and accuracy. In this dissertation, the proposed sampling-based approach is used to compute trajectories for an autonomous semi-buoyant solar-powered airship in the challenging Venusian atmosphere, which is characterized by super-rotation winds. A cost function that incorporates the energetic balance of the airship is proposed to find energy-efficient trajectories. This cost function combines the main forces acting on the vehicle: weight, buoyancy, aerodynamic lift and drag, and thrust. The FlowFMT* method is also extended to consider the possibility of battery depletion due to thrust or battery charging due to solar energy and tested in this Venus atmosphere scenario. Simulations showcase how the airship selects high-altitude paths to minimize energy consumption and maximize battery recharge. They also show the airship sinking down and drifting with the wind at the altitudes where it is fully buoyant. For terrestrial applications, this dissertation finally introduces the Sensor-Space Lattice (SSLAT) motion planner, a real-time obstacle avoidance algorithm for autonomous vehicles and mobile robots equipped with planar range finders. This planner uses a lattice to tessellate the area covered by the sensor and to rapidly compute collision-free paths in the robot surroundings by optimizing a cost function. The cost function guides the vehicle to follow an artificial vector field that encodes the desired vehicle path. This planner is evaluated in challenging, cluttered static environments, such as warehouses and forests, and in the presence of moving obstacles, both in simulations and real experiments. Our results show that our algorithm performs collision checking and path planning faster than baseline methods. Since the method can have sequential or parallel implementations, we also compare the two versions of SSLAT and show that the run-time for its parallel implementation, which is independent of the number and shape of the obstacles found in the environment, provides a significant speedup due to the independent collision checks.

Download

Share

COinS