Dynamic obstacle avoidance for quadrotors with event cameras

Abstract

Today’s autonomous drones have reaction times of tens of milliseconds, which is not enough for navigating fast in complex dynamic environments. To safely avoid fast moving objects, drones need low-latency sensors and algorithms. We departed from state-of-the-art approaches by using event cameras, which are bioinspired sensors with reaction times of microseconds. Our approach exploits the temporal information contained in the event stream to distinguish between static and dynamic objects and leverages a fast strategy to generate the motor commands necessary to avoid the approaching obstacles. Standard vision algorithms cannot be applied to event cameras because the output of these sensors is not images but a stream of asynchronous events that encode per-pixel intensity changes. Our resulting algorithm has an overall latency of only 3.5 milliseconds, which is sufficient for reliable detection and avoidance of fast-moving obstacles. We demonstrate the effectiveness of our approach on an autonomous quadrotor using only onboard sensing and computation. Our drone was capable of avoiding multiple obstacles of different sizes and shapes, at relative speeds up to 10 meters/second, both indoors and outdoors.

A Machine Learning Method for Vision-Based Unmanned Aerial Vehicle Systems to Understand Unknown Environments

“…Some methods use sensors in a passive way, such as event-based cameras [17], RGB-D cameras, and RGB cameras. These sensors are light in weight and rich in information.…”

“…These sensors are light in weight and rich in information. Falanga et al [17] used event-based cameras to achieve outdoor dynamic obstacle avoidance capability but the cost of the event-based camera is too high and it could not recognize generic items. Using RGB-D camera, researchers could generate a global map and compute the location easily, but these visual SLAM and VO methods require powerful computational capability, which is not suitable for UAVs with limited resources.…”

Auto-Tuned Event-Based Perception Scheme for Intrusion Monitoring With UAS

“…Then, a controller guided the UAS towards a safe direction opposite to the motion of the incoming obstacle. Recently, the work in [15] employed batches of events collected each 10 ms to perform ego-motion compensation for obstacle detection using a stereo event camera setup mounted in a quadrotor. The UAS performed a reactive avoidance strategy based on potential fields describing obstacles as geometric primitives.…”

“…The performance of some of the previous event-based methods, such as [18], [21], and [15], depend on finetuning their algorithm parameters. Meta-heuristic optimization strategies such as Simulated Annealing (SA) have been used for parameter tuning in computer vision and robotic applications, such as image segmentation [26], motion blur removal [27], feature selection [28], and robot path planning [29].…”

Weak Microbial Metabolites: a Treasure Trove for Using Biomimicry to Discover and Optimize Drugs

“…Biomimicry is derived from the words, bios (Greek) or life or nature, and mimesis (Greek) or imitation (Baumeister, 2012;Hargroves and Smith, 2006). Various industries have adopted biomimicry-based approaches for innovative solutions (Chauhan; Falanga et al, 2020;Hu et al, 2019;Wood, 2019). In medicine, biomimicry involves developing analogs of host endogenous molecules that have evolutionarily adapted to target a given receptor and induce a favorable outcome.…”