Sequential robot imitation learning from observations

This paper presents a framework to learn the sequential structure in the demonstrations for robot imitation learning. We first present a family of task-parameterized hidden semi-Markov models that extracts invariant segments (also called sub-goals or options) from demonstrated trajectories, and optimally follows the sampled sequence of states from the model with a linear quadratic tracking controller. We then extend the concept to learning invariant segments from visual observations that are sequenced together for robot imitation. We present Motion2Vec that learns a deep embedding space by minimizing a metric learning loss in a Siamese network: images from the same action segment are pulled together while being pushed away from randomly sampled images of other segments, and a time contrastive loss is used to preserve the temporal ordering of the images. The trained embeddings are segmented with a recurrent neural network, and subsequently used for decoding the end-effector pose of the robot. We first show its application to a pick-and-place task with the Baxter robot while avoiding a moving obstacle from four kinesthetic demonstrations only, followed by suturing task imitation from publicly available suturing videos of the JIGSAWS dataset with state-of-the-art [Formula: see text]% segmentation accuracy and [Formula: see text] cm error in position per observation on the test set.

Methods of Pre-Identification of TITO Systems

The content of this article is the presentation of methods used to identify systems before actual control, namely decentralized control of systems with Two Inputs, Two Outputs (TITO) and with two interactions. First, theoretical assumptions and reasons for using these methods are given. Subsequently, two methods for systems identification are described. At the end of this article, these specific methods are presented as the pre-identification of the chosen example. The Introduction part of the paper deals with the description of decentralized control, adaptive control, decentralized control in robotics and problem formulation (fixing the identification time at the existing decentralized self-tuning controller at the beginning of control and at the beginning of any set-point change) with the goal of a new method of identification. The Materials and methods section describes the used decentralized control method, recursive identification using approximation polynomials and least-squares with directional forgetting, recursive instrumental variable, self-tuning controller and suboptimal quadratic tracking controller, so all methods described in the section are those ones that already exist. Another section, named Assumptions, newly formulates the necessary background information, such as decentralized controllability and the system model, for the new identification method formulated in Pre-identification section. This section is followed by a section showing the results obtained by simulations and in real-time on a Coupled Drives model in the laboratory.

Two Inverse Problems Solution by Feedback Tracking Control

Two inverse ill-posed problems are considered. The first problem is an input restoration of a linear system. The second one is a restoration of time-dependent coefficients of a linear ordinary differential equation. Both problems are reformulated as auxiliary optimal control problems with regularizing cost functional. For the coefficients restoration problem, two control models are proposed. In the first model, the control coefficients are approximated by the output and the estimates of its derivatives. This model yields an approximating linear-quadratic optimal control problem having a known explicit solution. The derivatives are also obtained as auxiliary linear-quadratic tracking controls. The second control model is accurate and leads to a bilinear-quadratic optimal control problem. The latter is tackled in two ways: by an iterative procedure and by a feedback linearization. Simulation results show that a bilinear model provides more accurate coefficients estimates.

Autonomous Collision Avoidance Using MPC with LQR-Based Weight Transformation

Model predictive control (MPC) is a multi-objective control technique that can handle system constraints. However, the performance of an MPC controller highly relies on a proper prioritization weight for each objective, which highlights the need for a precise weight tuning technique. In this paper, we propose an analytical tuning technique by matching the MPC controller performance with the performance of a linear quadratic regulator (LQR) controller. The proposed methodology derives the transformation of a LQR weighting matrix with a fixed weighting factor using a discrete algebraic Riccati equation (DARE) and designs an MPC controller using the idea of a discrete time linear quadratic tracking problem (LQT) in the presence of constraints. The proposed methodology ensures optimal performance between unconstrained MPC and LQR controllers and provides a sub-optimal solution while the constraints are active during transient operations. The resulting MPC behaves as the discrete time LQR by selecting an appropriate weighting matrix in the MPC control problem and ensures the asymptotic stability of the system. In this paper, the effectiveness of the proposed technique is investigated in the application of a novel vehicle collision avoidance system that is designed in the form of linear inequality constraints within MPC. The simulation results confirm the potency of the proposed MPC control technique in performing a safe, feasible and collision-free path while respecting the inputs, states and collision avoidance constraints.

Sparse optimal control for a semilinear heat equation with mixed control-state constraints - regularity of Lagrange multipliers

An optimal control problem for a semilinear heat equation with distributed control is discussed, where two-sided pointwise box constraints on the control and two-sided pointwise mixed control-state constraints are given. The objective functional is the sum of a standard quadratic tracking type part and a multiple of the $L^1$-norm of the control that accounts for sparsity. Under a certain structural condition on almost active sets of the optimal solution, the existence of integrable Lagrange multipliers is proved for all inequality constraints. For this purpose, a theorem by Yosida and Hewitt is used. It is shown that the structural condition is fulfilled for all sufficiently large sparsity parameters. The sparsity of the optimal control is investigated. Eventually, higher smoothness of Lagrange multipliers is shown up to H\"older regularity.

Model reference adaptive LQT control for anti-jerk utilizing tire-road interaction characteristics

Sudden vehicle propulsion torque change under tip-in/out maneuver often leads to low-frequency longitudinal vibration due to the flexibility in the half-shaft and tire slip, which greatly affects vehicle drivability. Note that the vibration frequency is between 1 and 10 Hz and is difficult to be absorbed by the vehicle mechanical system. To optimize the vehicle drivability under tip-in maneuver, an Adaptive Linear Quadratic Tracking (ALQT) anti-jerk traction controller is proposed in this paper. Based on the experimental data, a Carsim-Simulink co-simulation model is developed for assessing control performance. A control-oriented model, considering the nonlinear characteristics of the tire-road friction coefficient and slip ratio, is then proposed. A reference model with rigid axle is used to provide the equilibrium points and reference velocity trajectory. Jacobi linearization method is then used to linearize the model along the desired trajectory and a linear deviation model based on equilibrium points is obtained. Finally, the deviation compensation receding horizon LQT controller is designed along with the Kalman state estimation. The effectiveness of the designed controller is assessed via simulation studies under different road surfaces and compared with PID and LQR controllers. The LQT controller is able to track the desired velocity profile with minimum jerk while increasing road safety. Furthermore, the effect of LQT weighting coefficients under different road surfaces are discussed. Simulation results show that the ALQT controller is able to optimize vehicle drivability under different road surfaces and the weighting matrices shall be selected based on the road condition for optimal drivability.