摘要：AbstractThis paper focuses on the comparison of control allocation (CA) techniques based on their application to a laboratory setup which emulates the rotational dynamics of a quadrotor. CA is an intermediate layer between controller and actuators which distributes the overall control effort in case of input redundancy. The CA algorithms Weighted Pseudoinverse, Redistributed Pseudoinverse, Enhanced Redistributed Pseudoinverse, Direct Allocation, Normalized Generalized Inverse, and Weighted Least Squares are briefly outlined and compared to each other. They are used to map the virtual torques generated by a second order sliding mode controller to the real actuators. Since in a real world application actuator power is always limited the performance of the methods in situations with multiple active constraints is investigated.
