def _compute_reward()

in gym_hil/envs/panda_pick_gym_env.py [0:0]

• 12 lines of code
• 2 McCabe index (conditional complexity)

    def _compute_reward(self) -> float:
        """Compute reward based on current state."""
        block_pos = self._data.sensor("block_pos").data

        if self.reward_type == "dense":
            tcp_pos = self._data.sensor("2f85/pinch_pos").data
            dist = np.linalg.norm(block_pos - tcp_pos)
            r_close = np.exp(-20 * dist)
            r_lift = (block_pos[2] - self._z_init) / (self._z_success - self._z_init)
            r_lift = np.clip(r_lift, 0.0, 1.0)
            return 0.3 * r_close + 0.7 * r_lift
        else:
            lift = block_pos[2] - self._z_init
            return float(lift > 0.1)