349cb5e0294fbba76e252bbd9026eb40a4651ae9,softlearning/environments/rllab/pusher_2d_env.py,ForkReacherEnv,reset,#ForkReacherEnv#Any#,232

Before Change

        qpos[self.JOINT_INDS[2]] = np.random.uniform(
            -np.pi/2, np.pi/2) + np.pi/2

        target_pos = np.random.uniform([-1.0], [1.0], size=[2])
        target_pos = np.sign(target_pos) * np.maximum(np.abs(target_pos), 1/2)
        target_pos[np.where(target_pos == 0)] = 1.0
        target_pos[1] += 1.0

        qpos[self.TARGET_INDS] = target_pos
        // qpos[self.TARGET_INDS] = [1.0, 2.0]
        // qpos[self.TARGET_INDS] = self.init_qpos.squeeze()[self.TARGET_INDS]

After Change

        qpos[self.JOINT_INDS[2]] = np.random.uniform(
            -np.pi/2, np.pi/2) + np.pi/2

        target_position = np.array(random_point_in_circle(
            angle_range=(0, 2*np.pi), radius=(0.6, 1.2)))
        target_position[1] += 1.0

        qpos[self.TARGET_INDS] = target_position
        // qpos[self.TARGET_INDS] = [1.0, 2.0]
        // qpos[self.TARGET_INDS] = self.init_qpos.squeeze()[self.TARGET_INDS]

Instances