gym/envs/mujoco/reacher.py [22:65]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
            self, "reacher.xml", 2, observation_space=observation_space, **kwargs
        )

    def step(self, a):
        vec = self.get_body_com("fingertip") - self.get_body_com("target")
        reward_dist = -np.linalg.norm(vec)
        reward_ctrl = -np.square(a).sum()
        reward = reward_dist + reward_ctrl

        self.do_simulation(a, self.frame_skip)
        if self.render_mode == "human":
            self.render()

        ob = self._get_obs()
        return (
            ob,
            reward,
            False,
            False,
            dict(reward_dist=reward_dist, reward_ctrl=reward_ctrl),
        )

    def viewer_setup(self):
        assert self.viewer is not None
        self.viewer.cam.trackbodyid = 0

    def reset_model(self):
        qpos = (
            self.np_random.uniform(low=-0.1, high=0.1, size=self.model.nq)
            + self.init_qpos
        )
        while True:
            self.goal = self.np_random.uniform(low=-0.2, high=0.2, size=2)
            if np.linalg.norm(self.goal) < 0.2:
                break
        qpos[-2:] = self.goal
        qvel = self.init_qvel + self.np_random.uniform(
            low=-0.005, high=0.005, size=self.model.nv
        )
        qvel[-2:] = 0
        self.set_state(qpos, qvel)
        return self._get_obs()

    def _get_obs(self):
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



gym/envs/mujoco/reacher_v4.py [134:177]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
            self, "reacher.xml", 2, observation_space=observation_space, **kwargs
        )

    def step(self, a):
        vec = self.get_body_com("fingertip") - self.get_body_com("target")
        reward_dist = -np.linalg.norm(vec)
        reward_ctrl = -np.square(a).sum()
        reward = reward_dist + reward_ctrl

        self.do_simulation(a, self.frame_skip)
        if self.render_mode == "human":
            self.render()

        ob = self._get_obs()
        return (
            ob,
            reward,
            False,
            False,
            dict(reward_dist=reward_dist, reward_ctrl=reward_ctrl),
        )

    def viewer_setup(self):
        assert self.viewer is not None
        self.viewer.cam.trackbodyid = 0

    def reset_model(self):
        qpos = (
            self.np_random.uniform(low=-0.1, high=0.1, size=self.model.nq)
            + self.init_qpos
        )
        while True:
            self.goal = self.np_random.uniform(low=-0.2, high=0.2, size=2)
            if np.linalg.norm(self.goal) < 0.2:
                break
        qpos[-2:] = self.goal
        qvel = self.init_qvel + self.np_random.uniform(
            low=-0.005, high=0.005, size=self.model.nv
        )
        qvel[-2:] = 0
        self.set_state(qpos, qvel)
        return self._get_obs()

    def _get_obs(self):
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -