Automatic re-initialization and failure recovery for aggressive flight with a monocular vision-based quadrotor

Abstract

Autonomous, vision-based quadrotor flight is widely regarded as a challenging perception and control problem since the accuracy of a flight maneuver is strongly influenced by the quality of the onboard state estimate. In addition, any vision-based state estimator can fail due to the lack of visual information in the scene or due to the loss of feature tracking after an aggressive maneuver. When this happens, the robot should automatically re-initialize the state estimate to maintain its autonomy and, thus, guarantee the safety for itself and the environment. In this paper, we present a system that enables a monocular-vision-based quadrotor to automatically recover from any unknown, initial attitude with significant velocity, such as after loss of visual tracking due to an aggressive maneuver. The recovery procedure consists of multiple stages, in which the quadrotor, first, stabilizes its attitude and altitude, then, re-initializes its visual state-estimation pipeline before stabilizing fully autonomously. To experimentally demonstrate the performance of our system, we aggressively throw the quadrotor in the air by hand and have it recover and stabilize all by itself. We chose this example as it simulates conditions similar to failure recovery during aggressive flight. Our system was able to recover successfully in several hundred throws in both indoor and outdoor environments. Abstract-Autonomous, vision-based quadrotor flight is widely regarded as a challenging perception and control problem since the accuracy of a flight maneuver is strongly influenced by the quality of the on-board state estimate. In addition, any vision-based state estimator can fail due to the lack of visual information in the scene or due to the loss of feature tracking after an aggressive maneuver. When this happens, the robot should automatically re-initialize the state estimate to maintain its autonomy and, thus, guarantee the safety for itself and the environment. In this paper, we present a system that enables a monocular-vision-based quadrotor to automatically recover from any unknown, initial attitude with significant velocity, such as after loss of visual tracking due to an aggressive maneuver. The recovery procedure consists of multiple stages, in which the quadrotor, first, stabilizes its attitude and altitude, then, re-initializes its visual state-estimation pipeline before stabilizing fully autonomously. To experimentally demonstrate the performance of our system, we aggressively throw the quadrotor in the air by hand and have it recover and stabilize all by itself. We chose this example as it simulates conditions similar to failure recovery during aggressive flight. Our system was able to recover successfully in several hundred throws in both indoor and outdoor environments.

Nonlinear ego-motion estimation from optical flow for online control of a quadrotor UAV

“…Acceleration measurements are then used to reconstruct the metric scale of the underlying map [6]. This has been achieved in [9], [10] using visual-SLAM algorithms, such as PTAM [11] and SVO [12] respectively, and, even more remarkably, through the observation of just a single feature over time by providing a closed-form solution for the computation of the metric scale factor [6]. However, all these approaches depend on the possibility to continuously track features for an extended period of time.…”

Simultaneous State Initialization and Gyroscope Bias Calibration in Visual Inertial Aided Navigation

“…For future work, we see this optimized closed-form solution being used on an MAV to provide accurate state initialization. This would allow aggressive take-off maneuvers, such as hand throwing the MAV in the air, as already demonstrated in [19] with a range sensor. With our technique, we could get rid of the range sensor.…”

“…A procedure to quickly re-initialize an MAV after a failure was presented in [19]. However, this method requires an altimeter to initialize the scale.…”

Vision-based autonomous quadrotor landing on a moving platform

“…The desired torques and the collective thrust are then converted to single motor thrusts. We refer the reader to our previous works [27] and [28] for further details on the dynamical model and the control algorithm used in this work.…”