Researchers from Stanford University and the University of Groningen in the United States have collaborated to enable a quadrotor UAV to grasp an aerial target during flight by introducing a passively triggered gripper. This research provides new ideas to enhance the ability of quadcopter drones to interact with their environment, which is expected to expand the range of applications for quadcopter drones.
For years, drones have been able to fly freely across the sky, but have not mastered the ability to land stably and often plant themselves. The ability of birds to wrap their talons around almost anything – pelicans skimming the tips of waves can suddenly land on pier pilings, and owls can swoop down at 64km/h to catch a mouse – has inspired researchers at Stanford University.
Researchers in the labs of Stanford engineers Mark Cutkosky and David Lentink are intrigued by the landing abilities of birds. Says Lentink, “For us, there was an idea that was inspiring – if you design different landing gear, robots can land anywhere.” So researchers from both labs worked together to develop an automated “robot claw” for “aerial robots” called SNAG, which approaches and lands on an object the same way every time, just like a real bird. SNAG was able to successfully grip beanbags and tennis balls in the air and release them in a controlled manner when triggered. The researchers have built on this functionality, so the drone has another aerial hunting skill to add to its repertoire!
How are Drone Claws Designed?
How exactly is it possible to catch a drone in flight with a steady and accurate grip? The first thing to be clear is that aerial capture involves three distinct phases: pre-collision pursuit of the target, capture, and then successful flight. The biggest challenge is the final step – successful flight after capture.
Previous long-legged bionic bird drones have been able to perform simple capture tasks, such as grabbing a small ball. In contrast to static objects, which are relatively simple because they cause little disturbance to the drone, capturing a moving object is a completely different matter, as the drone can easily be carried off course by the target, and can roll over if it is not careful.
To minimize interference and keep the UAV stable after capturing the target, the design of the claw is very important! The keywords are “light” and “fast”: reduced weight minimizes inertia, and a fast grip prevents the target from escaping! With this aim in mind, the researchers developed a passively triggered gripper mounted on a flexible suspension, which allows the gripper to close quickly using stored potential energy and is lighter than sensor monitoring. The result is a drone claw that weighs just 23 grams and closes quickly within 12 milliseconds of collision with a target!
In addition, the drone’s claws mimic the characteristics of a falcon: swinging its legs backward after catching its prey. This mechanism has two functions: the first is to reduce the relative velocity between the drone and the prey at the point of impact; the second is to move the prey’s center of gravity from in front of the drone to directly underneath it. This greatly increases stability after a successful capture.
Three Elements of Precise Control
The researchers constructed a model of the interaction between the UAV and the target, and based on the relative speeds of the two, they can calculate in real time the weight range of the target that can be safely captured, and also designed an automatically triggered gripper as well as a grapple controller.
The gripper weighs 23 g and is triggered by a passive mechanism without sensing and computational control loops, which minimizes the response time and can be closed within 12 ms; the grab controller consists of an on-board controller, a trajectory module, and a proportional-integral-derivative (PID) controller. Flight experiments show that a UAV weighing 550 g can grasp an airborne target weighing 85 g at a relative velocity in the range of 1 m/s~2.7 m/s. The controller is used for commanding and controlling the UAV.
- The onboard controller is used to command the motor rate to maintain the desired speed. It consists of an autopilot board, a companion computer, and a motor controller.
- The trajectory module calculates the straight-line trajectory between the start position, target position, and end position. To maintain a constant pitch throughout the flight, the module is designed to have a constant speed buffer before and after the target position.
- The PID controller module commands the grapple UAV to follow the desired trajectory. The controller obtains the position of the UAV and the desired position to calculate the required speed for the PX4 autopilot to follow. The low-level controller of the PX4 calculates the rotor rate required to fly at the commanded speed.
The control loop is shown in the figure. The trajectory module (orange) sends the desired position to the PID controller module, which then compares this value to the position determined by the mo-cap system (purple). This flight data (pink) is then sent between the modules. With this control scheme, precise control of the speed of the UAV upon impact with the target UAV was achieved.
This research won the IROS2022 Mechanisms and Design Best Paper Award and was published in the journal IEEE Robotics and Automation Letter under the title Aerial Grasping and the Velocity Sufficiency Region.
At the end of the paper, the researchers state that this research is a preliminary exploration of aerial hunting by drones and that there is still a lot of room for improvement in the future: for example, the “drone prey” can be visually tracked, to achieve comprehensive aerial tracking and capture; furthermore, the claw swinging mechanism can be further developed and tested to achieve higher relative speeds; and one way to improve is to increase the speed of the claw. Another enhancement would be to add an actively controlled joint at the base of the gripper or include more complex flight maneuvers to decouple the pitch and gripper mechanisms. Look forward to unlocking more skills in future drones!