SLAP: Slapband-based Autonomous Perching Drone with Failure Recovery for Vertical Tree Trunks

TL;DR

The paper addresses safe, energy-efficient perching of payload-bearing UAVs on vertical tree trunks and robust recovery from perch failures. It introduces SLAP, an integrated system combining a vision-based perch site detector, an IMU-based failure detector, an attitude controller for gentle perching, an optical close-range detector, and a fast active elastic gripper with microspines using slap bands. The approach emphasizes low-impact, forward-only perching with active grasping to reduce risk on larger drones, and includes a structured waterfall design process from simulation to indoor demonstration. Experimental results on a 1.2 kg quadrotor show 75% perch success across 20 HITL indoor flights and 100% recovery in two induced-failure tests, indicating promising safety and reliability for arboreal environmental monitoring and related tasks.

Abstract

Perching allows unmanned aerial vehicles (UAVs) to reduce energy consumption, remain anchored for surface sampling operations, or stably survey their surroundings. Previous efforts for perching on vertical surfaces have predominantly focused on lightweight mechanical design solutions with relatively scant system-level integration. Furthermore, perching strategies for vertical surfaces commonly require high-speed, aggressive landing operations that are dangerous for a surveyor drone with sensitive electronics onboard. This work presents the preliminary investigation of a perching approach suitable for larger drones that both gently perches on vertical tree trunks and reacts and recovers from perch failures. The system in this work, called SLAP, consists of vision-based perch site detector, an IMU (inertial-measurement-unit)-based perch failure detector, an attitude controller for soft perching, an optical close-range detection system, and a fast active elastic gripper with microspines made from commercially-available slapbands. We validated this approach on a modified 1.2 kg commercial quadrotor with component and system analysis. Initial human-in-the-loop autonomous indoor flight experiments achieved a 75% perch success rate on a real oak tree segment across 20 flights, and 100% perch failure recovery across 2 flights with induced failures.

Figures (12)

• Figure 1: (a) In mature forests, vertically-oriented tree trunks as highlighted in green are accessible areas for large drones to perch. (Photo taken by J. Di in Inyo National Forest.) Perching payload-capable drones could enable environmental monitoring of the surroundings. (b) A concept photograph of the perched drone system posed on a basswood tree. (c) A composite photo of a human-in-the-loop autonomous perching sequence of 1kg-class quadrotor on a real oak tree segment. The system identifies the perch site, flies a polynomial trajectory, and perches autonomously. The human verifies the perch site identified by the system before engaging perching.
• Figure 2: Autonomy state diagram. The system has human-in-the-loop check-ins to verify tree detection and to confirm that the system should start the perching maneuver. In the event of a perch failure, the system detects free fall and stabilizes itself in a hover at a safe distance from the ground and the tree.
• Figure 3: CAD rendering of the gripper design. A single servo is used to trigger the two friction latches through two tendons, and in turn, the latches release two bi-stable tape springs with microspines mounted at the tip.
• Figure 4: Vision pipeline overview. The raw RGBD camera images are displayed to the user, who prompts the tree search. The tree detector then detects and displays the bounding box and keypoints of a suitable perch site, which then are continuously tracked by the drone during flight and fed to the planner.
• Figure 5: Planner pipeline overview. The planner takes in tree state and drone state to generate polynomial trajectories to the target. The published outputs include the trajectory $P(t)$ and setpoints for the lower-level controller to follow.