Scale-aware navigation of a low-cost quadrocopter with a monocular camera

Abstract

We present a complete solution for the visual navigation of a small-scale, low-cost quadrocopter in unknown environments. Our approach relies solely on a monocular camera as the main sensor, and therefore does not need external tracking aids such as GPS or visual markers. Costly computations are carried out on an external laptop that communicates over wireless LAN with the quadrocopter. Our approach consists of three components: a monocular SLAM system, an extended Kalman filter for data fusion, and a PID controller. In this paper, we (1) propose a simple, yet effective method to compensate for large delays in the control loop using an accurate model of the quadrocopter's flight dynamics, and (2) present a novel, closed-form method to estimate the scale of a monocular SLAM system from additional metric sensors. We extensively evaluated our system in terms of pose estimation accuracy, flight accuracy, and flight agility using an external motion capture system. Furthermore, we compared the convergence and accuracy of our scale estimation method for an ultrasound altimeter and an air pressure sensor with filtering-based approaches. The complete system is available as open-source in ROS. This software can be used directly with a low-cost, off-the-shelf Parrot AR.Drone quadrocopter, and hence serves as an ideal basis for follow-up research projects.

Position Based Visual Servoing control of a Wheelchair Mounter Robotic Arm using Parallel Tracking and Mapping of task objects

“…Generated Pointcloudby PTAM However, the motion of the camera is not easily predictable, because of motion artifacts, robot dynamics uncertainly, friction, gearing backslash, etc. This nonlinear behavior makes the point cloud coordinates precise but not accurate, since the 3D points coordinates are scaled by an a-priori unknown factor λ [11]. In particular, we can model the PTAM measurements as Gaussian Random Variable with standard deviation σ P T AM and a mean λµ i , where µ i is the true position of the feature in the 3D space, and λ is an unknown scale factor.…”

“…It is possible to combine the output of the two methods by means of a Maximum-Likelihood Estimation method [11] in order to obtain an estimation of λ and correct the PTAM points' coordinates. This is equivalent to minimize the negative log-likelihood function for a given number n of acquisitions: …”

Development of a Control and Vision Interface for an AR.Drone

“…This is most noted in their use of the AR.Drone as a platform for control algorithms and vision navigation research, as well as a tool for educational and outreach purposes. Their research mainly pertains to the use of monocular RGB-D cameras to implement SLAM for the purpose of scale aware autonomous navigation and dense visual odometry [9]. In their research the vision group at TUM have employed a PID controller successfully on the Parrot Drone [9].…”

“…Their research mainly pertains to the use of monocular RGB-D cameras to implement SLAM for the purpose of scale aware autonomous navigation and dense visual odometry [9]. In their research the vision group at TUM have employed a PID controller successfully on the Parrot Drone [9]. Control techniques similar to those demonstrated by TUM have been shown in Altu & Taylor [10], as well as Bristeau et al [11].…”

“…The second camera points down, covers a field of view of 64 o and records at 60 fps. At any point in time only one video stream can be used to record video [9]. For movement the AR.Drone relies upon differential torque and thrust.…”