Planning and Control of Aggressive Maneuvers for Perching on Inclined and Vertical Surfaces

It is important to enable micro aerial vehicles to land and perch on different surfaces to save energy by cutting power to motors and to perform tasks such as persistent surveillance. In many cases, the best available surfaces may be vertical windows, walls, or inclined roof tops. In this paper, we present approaches and algorithms for aggressive maneuvering to enable perching of underactuated quadrotors on surfaces that are not horizontal. We show the design of a custom foot/gripper for perching on smooth surfaces. Then, we present control and planning algorithms for maneuvering to land on specified surfaces while satisfying constraints on actuation and sensing. Experimental results that include successful perching on vertical, glass surfaces validate the proposed techniques. INTRODUCTION Micro Aerial Vehicles (MAVs) have become ubiquitous, but they suffer from limited energy density batteries, which restricts ∗Address all correspondence to this author. †Department of Mechanical Engineering and Applied Mechanics their mission time [1]. Many tasks, however, do not require the robot to be in motion. In fact, some tasks do not even require the robot to be airborne. For example, a robot may be tasked with the objective to monitor a crime scene until police arrive or to monitor the gas levels of a nearby gas leak. In such cases, the vehicle could not only be stationary, but could also perch and turn off its motors to preserve energy for the next task. Perching capabilities could also be useful for tasks which require the robot to maintain a precise, static position, act as a radio relay in disaster zones, or to suspend operation during a period of unfavorable weather. Thus, perching is an appealing capability for aerial vehicles. Fixed wing vehicles are appealing because they typically have longer flight times than rotorcraft, but they cannot as easily hover in place. Multiple works leverage penetration-based grasping to perch using fixed-wing aircraft on vertical walls [2, 3, 4]. Other results enable perching on cables such as power lines [5,6]. However, detecting cables and performing relative pose estimation using onboard sensors and computation in real-time, realworld, scenarios would make such approaches very difficult in practice for fast-moving, micro-sized fixed-wing aircraft. Quadrotors, on the other hand, can carry larger payloads