Merge branch 'main' of github.com:MarkFzp/act-plus-plus into main

这个提交包含在:
Zipeng Fu
2023-10-26 01:44:36 -04:00
当前提交 8f22418a7b

查看文件

@ -198,7 +198,7 @@ def eval_bc(config, ckpt_name, save_episode=True):
if real_robot: if real_robot:
from aloha_scripts.robot_utils import move_grippers # requires aloha from aloha_scripts.robot_utils import move_grippers # requires aloha
from aloha_scripts.real_env import make_real_env # requires aloha from aloha_scripts.real_env import make_real_env # requires aloha
env = make_real_env(init_node=True, setup_base=True) env = make_real_env(init_node=True, setup_robots=True, setup_base=True)
env_max_reward = 0 env_max_reward = 0
else: else:
from sim_env import make_sim_env from sim_env import make_sim_env
@ -212,7 +212,7 @@ def eval_bc(config, ckpt_name, save_episode=True):
max_timesteps = int(max_timesteps * 1) # may increase for real-world tasks max_timesteps = int(max_timesteps * 1) # may increase for real-world tasks
num_rollouts = 50 num_rollouts = 1
episode_returns = [] episode_returns = []
highest_rewards = [] highest_rewards = []
for rollout_id in range(num_rollouts): for rollout_id in range(num_rollouts):
@ -233,7 +233,7 @@ def eval_bc(config, ckpt_name, save_episode=True):
### evaluation loop ### evaluation loop
if temporal_agg: if temporal_agg:
all_time_actions = torch.zeros([max_timesteps, max_timesteps+num_queries, state_dim]).cuda() all_time_actions = torch.zeros([max_timesteps, max_timesteps+num_queries, 16]).cuda()
qpos_history = torch.zeros((1, max_timesteps, state_dim)).cuda() qpos_history = torch.zeros((1, max_timesteps, state_dim)).cuda()
image_list = [] # for visualization image_list = [] # for visualization
@ -271,6 +271,7 @@ def eval_bc(config, ckpt_name, save_episode=True):
vq_sample = latent_model.generate(1, temperature=1, x=None) vq_sample = latent_model.generate(1, temperature=1, x=None)
all_actions = policy(qpos, curr_image, vq_sample=vq_sample) all_actions = policy(qpos, curr_image, vq_sample=vq_sample)
else: else:
# e()
all_actions = policy(qpos, curr_image) all_actions = policy(qpos, curr_image)
if temporal_agg: if temporal_agg:
all_time_actions[[t], t:t+num_queries] = all_actions all_time_actions[[t], t:t+num_queries] = all_actions