This is a list of the robo-gym environments.
For information on creating your own environment, see Creating Environments.
Available UR models: UR3, UR3e, UR5, UR5e, UR10, UR10e, UR16
To select the robot model use: ur_model='<ur3, ur3e, ur5, ur5e, ur10, ur10e, ur16e>'
General Warning: When resetting the environment, in some cases the robot moves to a random initial position. When executing the command to move the robot to the desired position we simply forward the random joint positions to the robot controller, a collision free path is not ensured. Therefore when using the Real Robot environment the robot could go in self collision during the reset stage, please be cautious and always keep the emergency stop at end when operating the real robot.
# simulated robot environment
env = gym.make('EmptyEnvironmentURSim-v0', ur_model='ur5', ip='<server_manager_address>')
# real robot environment
env = gym.make('EmptyEnvironmentURRob-v0', ur_model='ur5', rs_address='<robot_server_address>')
This is the base UR environment. This environment is not intended to be used as a standalone environment but rather as a starting point and base class to develop UR environments.
The environment state includes: joint positions normalized with respect to the joint position limits and the joint velocities (rad/s). The reward is constant to 0.
The robot uses position control; therefore, an action in the environment consists of six normalized joint position values.
# simulated robot environment
env = gym.make('EndEffectorPositioningURSim-v0', ur_model='ur10', ip='<server_manager_address>')
# real robot environment
env = gym.make('EndEffectorPositioningURRob-v0', ur_model='ur10', rs_address='<robot_server_address>')
The goal in this environment is for the robotic arm to reach a target position with its end effector.
The target end effector positions are uniformly distributed across a semi-sphere of the size close to the full working area of the robot. Potential target points generated within the singularity areas of the working space are discarded. The starting position is a random robot configuration.
The environment state includes: the 3D polar coordinates of the target position with respect to the end effector frame, joint positions normalized with respect to the joint position limits and the joint velocities (rad/s).
The robot uses position control; therefore, an action in the environment consists of normalized joint position values.
# simulated robot environment
env = gym.make('BasicAvoidanceURSim-v0', ur_model='ur5', ip='<server_manager_address>')
# real robot environment
env = gym.make('BasicAvoidanceURRob-v0', ur_model='ur5', rs_address='<robot_server_address>')
The goal in this environment is for the robotic arm to keep a minimum distance (calculated from end effector and the elbow) to an obstacle moving vertically while keeping as close as possible to the initial joint configuration.
The environment state includes: the 3D polar coordinates of the obstacle with respect to the end effector frame and with respect to the forearm link frame, joint positions normalized with respect to the joint position limits and the difference between the joint positions and the initial joint configuration.
An action in the environment consists in normalized joint position deltas from the initial joint configuration.
# simulated robot environment
env = gym.make('AvoidanceRaad2022URSim-v0', ur_model='ur5', ip='<server_manager_address>')
# real robot environment
env = gym.make('AvoidanceRaad2022URRob-v0', ur_model='ur5', rs_address='<robot_server_address>')
Environment used in our Paper Submission to RAAD 2022.
The goal in this environment is for the robotic arm to keep a minimum distance (calculated from end effector and the elbow) to an obstacle moving following 3D splines generated in the robot working area while keeping as close as possible to the pre-configured robot trajectory.
The environment state includes: the 3D polar coordinates of the obstacle with respect to the end effector frame and with respect to the forearm link frame, joint positions normalized with respect to the joint position limits, the difference between the joint positions and the trajectory joint configuration, the trajectory joint configuration and a flag to indicate whether the current trajectory joint configuration is a waypoint.
An action in the environment consists in normalized joint position deltas from the trajectory joint configuration.
# simulated robot environment
env = gym.make('AvoidanceRaad2022TestURSim-v0', ur_model='ur5', ip='<server_manager_address>')
# real robot environment
env = gym.make('AvoidanceRaad2022TestURRob-v0', ur_model='ur5', rs_address='<robot_server_address>')Same as Avoidance RAAD 2022 but using a fixed set of 3D splines as obstacles trajectories.
Available Interbotix arm models: ReactorX-150, PincherX-150, ReactorX-200, ViperX-250, ViperX-300, WidowX-200, WidowX-250, PincherX-100, ViperX-300S, WidowX-250S
To select the robot model use: robot_model='<rx150, px150, rx200, vx250, vx300, wx200, wx250, px100, vx300s, wx250s>'
General Warning: When resetting the environment, in some cases the robot moves to a random initial position. When executing the command to move the robot to the desired position we simply forward the random joint positions to the robot controller, a collision free path is not ensured. Therefore when using the Real Robot environment the robot could go in self collision during the reset stage, please be cautious and always keep the emergency stop at end when operating the real robot.
# simulated robot environment
env = gym.make('EmptyEnvironmentInterbotixASim-v0', robot_model='wx250s', ip='<server_manager_address>')
# real robot environment
env = gym.make('EmptyEnvironmentInterbotixARob-v0', robot_model='wx250s', rs_address='<robot_server_address>')This is the base Interbotix arm environment. This environment is not intended to be used as a standalone environment but rather as a starting point and base class to develop Interbotix arm environments.
The environment state includes: joint positions normalized with respect to the joint position limits and the joint velocities (rad/s). The reward is constant to 0.
The robot uses position control; therefore, an action in the environment consists of four, five or six (depending on robot model) normalized joint position values.
# simulated robot environment
env = gym.make('EndEffectorPositioningInterbotixASim-v0', robot_model='wx250s', ip='<server_manager_address>')
# real robot environment
env = gym.make('EndEffectorPositioningInterbotixARob-v0', robot_model='wx250s', rs_address='<robot_server_address>')The goal in this environment is for the robotic arm to reach a target position with its end effector.
The target end effector positions are uniformly distributed across a semi-sphere of the size close to the full working area of the robot. Potential target points generated within the singularity areas of the working space are discarded. The starting position is a random robot configuration.
The environment state includes: the 3D polar coordinates of the target position with respect to the end effector frame, joint positions normalized with respect to the joint position limits and the joint velocities (rad/s).
The robot uses position control; therefore, an action in the environment consists of normalized joint position values.
# simulated robot environment
env = gym.make('BasicAvoidanceInterbotixASim-v0', robot_model='wx250s', ip='<server_manager_address>')
# real robot environment
env = gym.make('BasicAvoidanceInterbotixARob-v0', robot_model='wx250s', rs_address='<robot_server_address>')The goal in this environment is for the robotic arm to keep a minimum distance (calculated from end effector and the elbow) to an obstacle moving vertically while keeping as close as possible to the initial joint configuration.
The environment state includes: the 3D polar coordinates of the obstacle with respect to the end effector frame and with respect to the forearm link frame, joint positions normalized with respect to the joint position limits and the difference between the joint positions and the initial joint configuration.
An action in the environment consists in normalized joint position deltas from the initial joint configuration.
# simulated robot environment
env = gym.make('NoObstacleNavigationMir100Sim-v0', ip='<server_manager_address>')
# real robot environment
env = gym.make('NoObstacleNavigationMir100Rob-v0', rs_address='<robot_server_address>')In this environment, the task of the mobile robot is to reach a target position in a obstacle-free environment. At the initialization of the environment the target is randomly generated within a 2x2m area. For the simulated environment the starting position of the robot is generated randomly whereas for the real robot the last robot's position is used.
The observations consist of 4 values. The first two are the polar coordinates of the target position in the robot's reference frame. The third and the fourth value are the linear and angular velocity of the robot.
The action is composed of two values: the target linear and angular velocity of the robot.
The base reward that the agent receives at each step is proportional to the variation of the two-dimensional Euclidean distance to the goal position. Thus, a positive reward is received for moving closer to the goal, whereas a negative reward is collected for moving away. In addition, the agent receives a large positive reward for reaching the goal and a large negative reward when crossing the external boundaries of the map.
# simulated robot environment
env = gym.make('ObstacleAvoidanceMir100Sim-v0', ip='<server_manager_address>')
# real robot environment
env = gym.make('ObstacleAvoidanceMir100Rob-v0', rs_address='<robot_server_address>')
In this environment, the task of the mobile robot is to reach a target position without touching the obstacles on the way. In order to detect obstacles, the MiR100 is equipped with two laser scanners, which provide distance measurements in all directions on a 2D plane. At the initialization of the environment the target is randomly placed on the opposite side of the map with respect to the robot's position. Furthermore, three cubes, which act as obstacles, are randomly placed in between the start and goal positions. The cubes have an edge length of 0.5 m, whereas the whole map measures 6x8 m. For the simulated environment the starting position of the robot is generated randomly whereas for the real robot the last robot's position is used.
The observations consist of 20 values. The first two are the polar coordinates of the target position in the robot's reference frame. The third and the fourth value are the linear and angular velocity of the robot. The remaining 16 are the distance measurements received from the laser scanner distributed evenly around the mobile robot. These values were downsampled from 2*501 laser scanner values to reduce the complexity of the learning task.
The action is composed of two values: the target linear and angular velocity of the robot.
The base reward that the agent receives at each step is proportional to the variation of the two-dimensional Euclidean distance to the goal position. Thus, a positive reward is received for moving closer to the goal, whereas a negative reward is collected for moving away. In addition, the agent receives a large positive reward for reaching the goal and a large negative reward in case of collision.