Robotics

AI helps drones find the perfect branch to perch on

Drones that can autonomously select the best tree branch to land on

Deep Dive

A team led by Alex Dunnett at Imperial College London has published a paper demonstrating a vision-based system that enables drones to autonomously identify the best branch to perch on in a tree. Instead of simply targeting the nearest available branch, the system analyzes each branch's shape and structure using image processing, machine learning, image segmentation, and binary morphology. It assesses branch width, slope (angle to horizontal), and curvature to determine suitability for a tendon-driven grasping claw.

The method was validated on a dataset of over 10,000 urban tree images taken from February to October in subtropical and temperate monsoon climates. For feasible targets (branches thick enough and with sufficient perching space), the system successfully located a perch spot 76% of the time. The researchers note this is a preliminary result and plan to incorporate depth perception and attitude sensors to improve performance. The work could lead to drones that can recharge, survey, or monitor environments by perching in trees.

Key Points
  • Vision-guided system evaluates branch width, slope, and curvature for optimal perching
  • Tested on 10,000+ urban tree images with 76% success on feasible targets
  • Uses machine learning, segmentation, and morphology to assess branch suitability

Why It Matters

Enables autonomous drones to perch in trees for monitoring, inspection, and recharging in natural environments