Search

A new method for enabling a quadrotor micro air vehicle (MAV) to navigate unknown environments using reinforcement learning (RL) and model predictive control (MPC) is developed. An efficient implementation of MPC provides vehicle control and obstacle avoidance. RL is used to guide the MAV through complex environments where dead-end corridors may be encountered and backtracking is necessary. All of the presented algorithms were deployed on embedded hardware using automatic code generation from Simulink. Results are given for flight tests, demonstrating that the algorithms perform well with modest computing requirements and robust navigation.

Assistive teleoperation aims to help operators control robotic systems with ease. In this work, we present a novel adaptive teleoperation approach that is amenable to mobile systems using motion primitives for long-duration teleoperation, such as exploration using mobile vehicles or walking for humanoid systems. We first describe teleoperation using motion primitives, which are dynamically feasible and safe local trajectories based on a kinematic or dynamic model. We take a predict-and-adapt approach to assistive teleoperation, whereby adaptation is based on the predicted user intent. By representing the operator as an optimizing controller, a probabilistic distribution can be constructed for the available future actions based on some reward function. Adaptation is provided in the form of subsampling, which tailors the set of available actions based on the likelihood of action selection. We describe the framework for general systems and delineate the extrapolation to ground, air, and legged mobile robots, and demonstrate generalizability of this framework on two systems via simulation and experimentation; namely, a quadrotor micro air vehicle, and a simulated 3D humanoid system. Both systems show provably better performance in teleoperation by measures of behavioral entropy.

This paper presents a vision-based collision avoidance technique for small and miniature air vehicles (MAVs) using local-level frame mapping and path planning. Using computer vision algorithms, a depth map that represents the range and bearing to obstacles is obtained. Based on the depth map, we estimate the range, azimuth to, and height of obstacles using an extended Kalman filter that takes into account the correlations between obstacles. We then construct maps in the local-level frame using cylindrical coordinates for three dimensional path planning and plan Dubins paths using the rapidly-exploring random tree algorithm. The behavior of our approach is analyzed and the characteristics of the environments where the local path planning technique guarantees collision-free paths and maneuvers the MAV to a specific goal region are described. Numerical results show the proposed technique is successful in solving path planning and multiple obstacle avoidance problems for fixed wing MAVs.

The problem of developing a reliable system for sensing and controlling the hover of a Micro Air Vehicle (MAV) using visual snapshots is considered. The current problem is part of a larger project, which is developing an autonomous MAV, controlled by vision only information. A new algorithm is proposed that uses a stored image of the ground, a snapshot taken of the ground directly under the MAV, as a visual anchor point. The absolute translation of the aircraft and its velocity are then calculated by comparing the subsequent frames with the stored image and fed into the position controller. In order to increase the performance, several issues, such as effects of scale uncertainty on the closed loop stability of the platform are investigated. For controller design and testing purposes, we analytically derive a complete model of a small size helicopter with no stabilizing bar (flybar). The simulation results for 2D and 3D snapshots confirm the effectiveness of the proposed algorithm.

This paper presents the design and development of autonomous attitude stabilization, navigation in unstructured, GPS-denied environments, aggressive landing on inclined surfaces, and aerial gripping using onboard sensors on a low-cost, custom-built quadrotor. The development of a multi-functional micro air vehicle (MAV) that utilizes inexpensive off-the-shelf components presents multiple challenges due to noise and sensor accuracy, and there are control challenges involved with achieving various capabilities beyond navigation. This paper addresses these issues by developing a complete system from the ground up, addressing the attitude stabilization problem using extensive filtering and an attitude estimation filter recently developed in the literature. Navigation in both indoor and outdoor environments is achieved using a visual Simultaneous Localization and Mapping (SLAM) algorithm that relies on an onboard monocular camera. The system utilizes nested controllers for attitude stabilization, vision-based navigation, and guidance, with the navigation controller implemented using a nonlinear controller based on the sigmoid function. The efficacy of the approach is demonstrated by maintaining a stable hover even in the presence of wind gusts and when manually hitting and pulling on the quadrotor. Precision landing on inclined surfaces is demonstrated as an example of an aggressive maneuver, and is performed using only onboard sensing. Aerial gripping is accomplished with the addition of a secondary camera, capable of detecting infrared light sources, which is used to estimate the 3D location of an object, while an under-actuated and passively compliant manipulator is designed for effective gripping under uncertainty. The quadrotor is therefore able to autonomously navigate inside and outside, in the presence of disturbances, and perform tasks such as aggressively landing on inclined surfaces and locating and grasping an object, using only inexpensive, onboard sensors.

Insects are capable of robust visual navigation in complex environments using efficient information extraction and processing approaches. This paper presents an implementation of insect inspired visual navigation that uses spatial decompositions of the instantaneous optic flow to extract local proximity information. The approach is demonstrated in a corridor environment on an autonomous quadrotor micro-air-vehicle (MAV) where all the sensing and processing, including altitude, attitude, and outer loop control is performed on-board. The resulting methodology has the advantages of computation speed and simplicity, hence are consistent with the stringent size, weight, and power requirements of MAVs.