Real-time control for robotics is a popular research area in the reinforcement learning community. Through the use of techniques such as reward shaping, researchers have managed to train online agents across a multitude of domains. Despite these advances, solving goal-oriented tasks still requires complex architectural changes or hard constraints to be placed on the problem. In this article, we solve the problem of stacking multiple cubes by combining curriculum learning, reward shaping, and a high number of efficiently parallelized environments. We introduce two curriculum learning settings that allow us to separate the complex task into sequential sub-goals, hence enabling the learning of a problem that may otherwise be too difficult. We focus on discussing the challenges encountered while implementing them in a goal-conditioned environment. Finally, we extend the best configuration identified on a higher complexity environment with differently shaped objects.
One of the major benefits of quantum computing is the potential to resolve complex computational problems faster than can be done by classical methods. There are many prototype-based clustering methods in use today, and the selection of the starting nodes for the center points is often done randomly. Clustering often suffers from accepting a local minima as a valid solution when there are possibly better solutions. We will present the results of a study to leverage the benefits of quantum computing for finding better starting centroids for prototype-based clustering.
Simulation-to-real is the task of training and developing machine learning models and deploying them in real settings with minimal additional training. This approach is becoming increasingly popular in fields such as robotics. However, there is often a gap between the simulated environment and the real world, and machine learning models trained in simulation may not perform as well in the real world. We propose a framework that utilizes a message-passing pipeline to minimize the information gap between simulation and reality. The message-passing pipeline is comprised of three modules: scene understanding, robot planning, and performance validation. First, the scene understanding module aims to match the scene layout between the real environment set-up and its digital twin. Then, the robot planning module solves a robotic task through trial and error in the simulation. Finally, the performance validation module varies the planning results by constantly checking the status difference of the robot and object status between the real set-up and the simulation. In the experiment, we perform a case study that requires a robot to make a cup of coffee. Results show that the robot is able to complete the task under our framework successfully. The robot follows the steps programmed into its system and utilizes its actuators to interact with the coffee machine and other tools required for the task. A noteworthy observation from the experiment is the speed and accuracy with which the robot completed the task. The robot can make a cup of coffee relatively quickly compared with traditional robot planning and control methods, and its movements were precise and efficient. Overall, the results of this case study demonstrate the potential benefits of our method that drive robots for tasks that require precision and efficiency. Further research in this area could lead to the development of even more versatile and adaptable robots, opening up new possibilities for automation in various industries.
In this paper, a position-based visual servo control scheme for a bottle packaging process using the ScorBot-ER-4U robot manipulator is proposed. The control scheme considers the eye-to-hand configuration and is based on the kinematic model of the manipulator robot and the perspective projection model of the vision camera (Pinhole Model). The proposed scheme was evaluated on a virtual environment developed in the Unity3D graphics engine; and experimentally with the ScorBot-ER-4U manipulator robot and the ZED 2 stereo vision camera. Finally, it is concluded that the results obtained by simulation and experimentally show that the control errors converge to zero asymptotically.
Vehicle classification is an essential part of intelligent transportation systems (ITS). This work proposes a model based on transfer learning, combining data augmentation for the recognition and classification of local vehicle classes in Canada. It takes inspiration from contemporary deep learning (DL) achievements for image classification. This makes use of the Dataset named Stanford AI of Vehicles, which has 16185 images. The images in this section are divided into 196 types of vehicles. To increase performance further, additional classification blocks are added to the residual network (ResNet-50)-based model which is being used. In this case, vehicle type details are automatically extracted and classified. A number of measures like accuracy, precision, recall, etc. are used during the analysis to evaluate the results. The proposed model exhibits increased accuracy despite vehicles’ different physical characteristics. In comparison to the current baseline method and the two pre-trained DL systems, AlexNet and VGG-16, our suggested method outperforms them all. The suggested ResNet-50 pre-trained model achieves an accuracy of 90.07
