ABSTRACT
To enable the design of planning and control strategies in simulated environments before their direct application to the real robot, exploiting the Sim2Real practice, powerful and realistic dynamic simulation tools have been proposed, e.g., the ROS-Gazebo framework. However, the majority of such simulators do not account for some of the properties of recently developed advanced systems, e.g., dynamic elastic behaviors shown by all those robots that purposely incorporate compliant elements into their actuators, the so-called Articulated Soft Robots ASRs. This paper presents an open-source ROS-Gazebo toolbox for simulating ASRs equipped with the aforementioned types of compliant actuators. To achieve this result, the toolbox consists of two ROS-Gazebo modules: a plugin that implements the custom compliant characteristics of a given actuator and simulates the internal motor dynamics, and a Robotic Operation System (ROS) manager node used to organize and simplify the overall toolbox usage. The toolbox can implement different compliant joint structures to perform realistic and representative simulations of ASRs, also when they interact with the environment. The simulated ASRs can be also used to retrieve information about the physical behavior of the real system from its simulation, and to develop control policies that can be transferred back to the real world, leveraging the Sim2Real practice. To assess the versatility of the proposed plugin, we report simulations of different compliant actuators. Then, to show the reliability of the simulated results, we present experiments executed on two ASRs and compare the performance of the real hardware with the simulations. Finally, to validate the toolbox effectiveness for Sim2Real control design, we learn a control policy in simulation, then feed it to the real system in feed-forward comparing the results.
ABSTRACT
Self driving vehicles promise to bring one of the greatest technological and social revolutions of the next decade for their potential to drastically change human mobility and goods transportation, in particular regarding efficiency and safety. Autonomous racing provides very similar technological issues while allowing for more extreme conditions in a safe human environment. While the software stack driving the racing car consists of several modules, in this paper we focus on the localization problem, which provides as output the estimated pose of the vehicle needed by the planning and control modules. When driving near the friction limits, localization accuracy is critical as small errors can induce large errors in control due to the nonlinearities of the vehicle's dynamic model. In this paper, we present a localization architecture for a racing car that does not rely on Global Navigation Satellite Systems (GNSS). It consists of two multi-rate Extended Kalman Filters and an extension of a state-of-the-art laser-based Monte Carlo localization approach that exploits some a priori knowledge of the environment and context. We first compare the proposed method with a solution based on a widely employed state-of-the-art implementation, outlining its strengths and limitations within our experimental scenario. The architecture is then tested both in simulation and experimentally on a full-scale autonomous electric racing car during an event of Roborace Season Alpha. The results show its robustness in avoiding the robot kidnapping problem typical of particle filters localization methods, while providing a smooth and high rate pose estimate. The pose error distribution depends on the car velocity, and spans on average from 0.1 m (at 60 km/h) to 1.48 m (at 200 km/h) laterally and from 1.9 m (at 100 km/h) to 4.92 m (at 200 km/h) longitudinally.
