Vision-Based Distributed Formation Control of Unmanned Aerial Vehicles

We present a novel control strategy for a team of unmanned aerial vehicles (UAVs) to autonomously achieve a desired formation using only visual feedback provided by the UAV's onboard cameras. This effectively eliminates the need for global position measurements. The proposed pipeline is fully distributed and encompasses a collision avoidance scheme. In our approach, each UAV extracts feature points from captured images and communicates their pixel coordinates and descriptors among its neighbors. These feature points are used in our novel pose estimation algorithm, QuEst, to localize the neighboring UAVs. Compared to existing methods, QuEst has better estimation accuracy and is robust to feature point degeneracies. We demonstrate the proposed pipeline in a high-fidelity simulation environment and show that UAVs can achieve a desired formation in a natural environment without any fiducial markers.

Robust 3D Distributed Formation Control with Application to Quadrotors

We present a distributed control strategy for a team of quadrotors to autonomously achieve a desired 3D formation. Our approach is based on local relative position measurements and does not require global position information or inter-vehicle communication. We assume that quadrotors have a common sense of direction, which is chosen as the direction of gravitational force measured by their onboard IMU sensors. However, this assumption is not crucial, and our approach is robust to inaccuracies and effects of acceleration on gravitational measurements. In particular, converge to the desired formation is unaffected if each quadrotor has a velocity vector that projects positively onto the desired velocity vector provided by the formation control strategy. We demonstrate the validity of proposed approach in an experimental setup and show that a team of quadrotors achieve a desired 3D formation.

Robust Distributed Planar Formation Control for Higher-Order Holonomic and Nonholonomic Agents

We present a distributed formation control strategy for agents with a variety of dynamics to achieve a desired planar formation. The proposed strategy is fully distributed, does not require inter-agent communication or a common sense of orientation, and can be implemented using relative position measurements acquired by agents in their local coordinate frames. We show how the control designed for agents with the simplest dynamical model, i.e., the single-integrator dynamics, can be extended to holonomic agents with higher-order dynamics such as quadrotors, and nonholonomic agents such as unicycles and cars. We prove that the proposed strategy is robust to saturations in the input, unmodeled dynamics, and switches in the sensing topology. We further show that the control is relaxed in the sense that agents can move along a rotated and scaled control direction without affecting the convergence to the desired formation. This observation is used to design a distributed collision avoidance strategy. We demonstrate the proposed approach in simulations and further present a distributed robotic platform to test the strategy experimentally. Our experimental platform consists of off-the-shelf equipment that can be used to test and validate other multi-agent algorithms. The code and implementation instructions for this platform are available online and free.

Quaternion Based Camera Pose Estimation From Matched Feature Points

We present a novel solution to the camera pose estimation problem, where rotation and translation of a camera between two views are estimated from matched feature points in the images. The camera pose estimation problem is traditionally solved via algorithms that are based on the essential matrix or the Euclidean homography. With six or more feature points in general positions in the space, essential matrix based algorithms can recover a unique solution. However, such algorithms fail when points are on critical surfaces (e.g., coplanar points) and homography should be used instead. By formulating the problem in quaternions and decoupling the rotation and translation estimation, our proposed algorithm works for all point configurations. Using both simulated and real world images, we compare the estimation accuracy of our algorithm with some of the most commonly used algorithms. Our method is shown to be more robust to noise and outliers. For the benefit of community, we have made the implementation of our algorithm available online and free.

Vision-based Control of a Soft Robot for Maskless Head and Neck Cancer Radiotherapy

This work presents an on-going investigation of the control of a pneumatic soft-robot actuator addressing accurate patient positioning systems in maskless head and neck cancer radiotherapy. We employ two RGB-D sensors in a sensor fusion scheme to better estimate a patient's head pitch motion. A system identification prediction error model is used to obtain a linear time invariant state space model. We then use the model to design a linear quadratic Gaussian feedback controller to manipulate the patient head position based on sensed head pitch motion. Experiments demonstrate the success of our approach.