Search

In this paper, the solvability of the Inverse Optimal Control (IOC) problem based on two existing minimum principal methods, is analysed. The aim of this work is to answer the question regarding what kinds of trajectories, that is depending on the initial conditions of the closed-loop system and system dynamics, of the original optimal control problem, will result in the recovery of the true weights of the reward function for both the soft and the hard-constrained methods [1], [2]. Analytical conditions are provided which allow to verify if a trajectory is sufficiently conditioned, that is, holds sufficient information to recover the true weights of an optimal control problem. It was found that the open-loop system of the original optimal problem has a stronger influence on the solvability of the Inverse Optimal Control problem for the hard-constrained method as compared to the soft-constrained method. These analytical results were validated via simulation., Comment: 8 pages, submitted to IEEE Transactions on Automatic Control

In harsh environments such as those found in nuclear facilities, the use of robotic systems is crucial for performing tasks that would otherwise require human intervention. This is done to minimize the risk of human exposure to dangerous levels of radiation, which can have severe consequences for health and even be fatal. However, the telemanipulation systems employed in these environments are becoming increasingly intricate, relying heavily on sophisticated control methods and local master devices. Consequently, the cognitive burden on operators during labor-intensive tasks is growing. To tackle this challenge, operator intention detection based on task learning can greatly enhance the performance of robotic tasks while reducing the reliance on human effort in teleoperation, particularly in a glovebox environment. By accurately predicting the operator's intentions, the robot can carry out tasks more efficiently and effectively, with minimal input from the operator. In this regard, we propose the utilization of Convolutional Neural Networks, a machine learning approach, to learn and forecast the operator's intentions using raw force feedback spatiotemporal data. Through our experimental study on glovebox tasks for nuclear applications, such as radiation survey and object grasping, we have achieved promising outcomes. Our approach holds the potential to enhance the safety and efficiency of robotic systems in harsh environments, thus diminishing the risk of human exposure to radiation while simultaneously improving the precision and speed of robotic operations.

Sim-and-real training is a promising alternative to sim-to-real training for robot manipulations. However, the current sim-and-real training is neither efficient, i.e., slow convergence to the optimal policy, nor effective, i.e., sizeable real-world robot data. Given limited time and hardware budgets, the performance of sim-and-real training is not satisfactory. In this paper, we propose a Consensus-based Sim-And-Real deep reinforcement learning algorithm (CSAR) for manipulator pick-and-place tasks, which shows comparable performance in both sim-and-real worlds. In this algorithm, we train the agents in simulators and the real world to get the optimal policies for both sim-and-real worlds. We found two interesting phenomenons: (1) Best policy in simulation is not the best for sim-and-real training. (2) The more simulation agents, the better sim-and-real training. The experimental video is available at: https://youtu.be/mcHJtNIsTEQ., Comment: 7 pages, 8 figures, IEEE International Conference on Robotics and Automation (ICRA) 2023

As autonomous machines such as robots and vehicles start performing tasks involving human users, ensuring a safe interaction between them becomes an important issue. Translating methods from human-robot interaction (HRI) studies to the interaction between humans and other highly complex machines (e.g. semi-autonomous vehicles) could help advance the use of those machines in scenarios requiring human interaction. One method involves understanding human intentions and decision-making to estimate the human's present and near-future actions whilst interacting with a robot. This idea originates from the psychological concept of Theory of Mind, which has been broadly explored for robotics and recently for autonomous and semi-autonomous vehicles. In this work, we explored how to predict human intentions before an action is performed by combining data from human-motion, vehicle-state and human inputs (e.g. steering wheel, pedals). A data-driven approach based on Recurrent Neural Network models was used to classify the current driving manoeuvre and to predict the future manoeuvre to be performed. A state-transition model was used with a fixed set of manoeuvres to label data recorded during the trials for real-time applications. Models were trained and tested using drivers of different seat preferences, driving expertise and arm-length; precision and recall metrics over 95% for manoeuvre identification and 86% for manoeuvre prediction were achieved, with prediction time-windows of up to 1 second for both known and unknown test subjects. Compared to our previous results, performance improved and manoeuvre prediction was possible for unknown test subjects without knowing the current manoeuvre., Comment: Submitted to IEEE Transactions on Human-Machine Systems on 17-Jul-2020

We show that a network can self-organize its structure in a completely distributed manner in order to optimize its synchronizability whilst satisfying the local constraints: non-negativity of edge weights, and maximum weighted degree of nodes. A novel multilayer approach is presented which uses a distributed strategy to estimate two spectral functions of the graph Laplacian, the algebraic connectivity $\lambda_2$ and the eigenratio $r = \lambda_n / \lambda_2$ . These local estimates are then used to evolve the edge weights so as to maximize $\lambda_2$, or minimize $r$ and, hence, achieve an optimal structure.