BiobotGrammar/env.py at main · benquick123/BiobotGrammar · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
import numpy as np
from utils import apply_action_clipping_sim


class SimEnvWrapper:

    def __init__(self, make_sim_and_task_fn, load=False):
        self.env, self.task = make_sim_and_task_fn()

        if load:
            self.env.restore_state_from_file("tmp.bullet")
        else:
            self.env.save_state_to_file("tmp.bullet")

        self.observation_dim = (None, )
        self.action_dim = self.env.get_robot_dof_count(0)
        self.seed = None
        self.real_step = False
        self.sim_joint_positions = np.zeros(self.action_dim)

    def step(self, action, torques=None):
        r = 0

        if torques is not None:
            self.env.get_joint_positions(0, self.sim_joint_positions)
            _, over_limits = apply_action_clipping_sim(self.sim_joint_positions, return_over_limit=True)
            torques = torques[:len(over_limits)]
            torques[over_limits] = 1

            # torques = np.zeros_like(torques)
            self.env.update_torques(0, torques)

        for k in range(self.task.interval):
            self.env.set_joint_targets(0, action.reshape(-1, 1))
            self.task.add_noise(self.env, (self.task.interval * self.seed + k) % (2 ** 32))
            self.env.step()

            r += self.task.get_objective_fn()(self.env) # , neural_input)

        if self.real_step:
            self.env.save_state_to_file("tmp.bullet")

        return None, r, None, None

    def reset(self, seed=None):
        if self.seed is None and seed is not None:
            self.seed = seed

    def set_seed(self, seed=None):
        self.seed = seed

    def set_env_state(self):
        self.env.restore_state()

    def real_env_step(self, is_real):
        self.real_step = is_real
        if not is_real:
            self.env.save_state()