tactile_tasks/eval_eva.py

#!/usr/bin/env python3
"""
Evaluate trained EVA policy in robosuite environment.

Same structure as eval_act.py, but adds tactile observation.

Usage:
  python tactile_tasks/eval_eva.py --task peg_insertion
  python tactile_tasks/eval_eva.py --task peg_insertion --temporal_agg
  python tactile_tasks/eval_eva.py --task peg_insertion --ckpt policy_last.ckpt
"""

import os
import sys
import argparse
import pickle

import numpy as np
import torch

# Setup paths
sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..", "policy", "EVA"))
sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..", "policy", "EVA", "detr"))
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))

from eva_policy import EVAPolicy
from tactile_tasks.uskin_sensor import USkinSensor


def load_policy(ckpt_dir, ckpt_name, policy_config):
    """Load trained EVA policy and dataset stats."""
    policy = EVAPolicy(policy_config)
    ckpt_path = os.path.join(ckpt_dir, ckpt_name)
    loading_status = policy.load_state_dict(torch.load(ckpt_path, map_location="cpu"))
    print(f"Loaded: {ckpt_path} ({loading_status})")
    policy.cuda()
    policy.eval()

    stats_path = os.path.join(ckpt_dir, "dataset_stats.pkl")
    with open(stats_path, "rb") as f:
        stats = pickle.load(f)

    return policy, stats


def get_obs(env, camera_names, stats):
    """Extract qpos and images from robosuite env, apply normalization."""
    robot = env.robots[0]

    # qpos: joint_pos(7) + normalized_gripper(1) = 8D
    joint_pos = np.array(env.sim.data.qpos[robot._ref_joint_pos_indexes])
    gripper_idx = robot._ref_gripper_joint_pos_indexes.get("right", [])
    gripper_val = env.sim.data.qpos[gripper_idx][0] / 0.8 if len(gripper_idx) else 0.0
    qpos = np.concatenate([joint_pos, [gripper_val]]).astype(np.float32)

    # Normalize qpos
    qpos_norm = (qpos - stats["qpos_mean"]) / stats["qpos_std"]
    qpos_tensor = torch.from_numpy(qpos_norm).float().cuda().unsqueeze(0)

    # Images: stack cameras -> (1, num_cams, 3, H, W)
    obs = env._get_observations()
    cam_map = {"agentview": "agentview_image", "eye_in_hand": "robot0_eye_in_hand_image"}
    images = []
    for cam in camera_names:
        key = cam_map.get(cam, cam + "_image")
        img = obs[key]  # (H, W, 3) uint8
        img = np.transpose(img, (2, 0, 1))  # -> (3, H, W)
        images.append(img)
    images = np.stack(images, axis=0)  # (num_cams, 3, H, W)
    images = torch.from_numpy(images / 255.0).float().cuda().unsqueeze(0)

    return qpos_tensor, images


def get_tactile(sensor):
    """Read one control-step worth of tactile data: [1, 2, 5, 4, 4, 3]."""
    left_frames = []
    right_frames = []
    for _ in range(USkinSensor.FREQ_MULTIPLIER):
        sensor.update()
        left_frames.append(sensor._left_forces.copy())
        right_frames.append(sensor._right_forces.copy())
    tl = np.stack(left_frames)   # [5, 4, 4, 3]
    tr = np.stack(right_frames)  # [5, 4, 4, 3]
    tactile = np.stack([tl, tr])  # [2, 5, 4, 4, 3]
    return torch.from_numpy(tactile).float().cuda().unsqueeze(0)


def run_eval(task, ckpt_dir, ckpt_name, num_rollouts, temporal_agg, camera_names,
             render=False):
    """Run evaluation rollouts."""
    import yaml

    # Load EVA config
    config_path = os.path.join(os.path.dirname(__file__), "..", "policy", "EVA", "train_config.yaml")
    with open(config_path, "r") as f:
        cfg = yaml.safe_load(f)

    state_dim = cfg["state_dim"]
    chunk_size = cfg["chunk_size"]

    policy_config = {
        "lr": cfg["lr"],
        "num_queries": chunk_size,
        "kl_weight": cfg["kl_weight"],
        "event_weight": cfg.get("event_weight", 0.1),
        "hidden_dim": cfg["hidden_dim"],
        "dim_feedforward": cfg["dim_feedforward"],
        "lr_backbone": 1e-5,
        "backbone": "resnet18",
        "enc_layers": 4,
        "dec_layers": 7,
        "nheads": 8,
        "camera_names": camera_names,
        "state_dim": state_dim,
        "chunk_size": chunk_size,
    }

    policy, stats = load_policy(ckpt_dir, ckpt_name, policy_config)
    post_process = lambda a: a * stats["action_std"] + stats["action_mean"]

    # Create environment
    from tactile_tasks.collect_data import create_env, TASK_CONFIGS
    settle_steps = 50
    max_timesteps = max(TASK_CONFIGS[task]["horizon"], 900)
    env = create_env(task, has_renderer=render)
    env.horizon = max_timesteps + settle_steps

    sensor = USkinSensor(env.sim)
    query_frequency = 1 if temporal_agg else chunk_size

    successes = []
    for ep in range(num_rollouts):
        env.reset()
        for _ in range(settle_steps):
            env.step(np.zeros(7))

        if temporal_agg:
            all_time_actions = torch.zeros(
                [max_timesteps, max_timesteps + chunk_size, state_dim]
            ).cuda()

        all_actions = None
        episode_success = False

        with torch.inference_mode():
            for t in range(max_timesteps):
                qpos, images = get_obs(env, camera_names, stats)
                tactile = get_tactile(sensor)

                # Query policy
                if t % query_frequency == 0:
                    all_actions = policy(qpos, images, tactile)  # (1, chunk_size, state_dim)

                if temporal_agg:
                    all_time_actions[[t], t:t + chunk_size] = all_actions
                    actions_for_curr_step = all_time_actions[:, t]
                    actions_populated = torch.all(actions_for_curr_step != 0, axis=1)
                    actions_for_curr_step = actions_for_curr_step[actions_populated]
                    k = 0.01
                    exp_weights = np.exp(-k * np.arange(len(actions_for_curr_step)))
                    exp_weights = exp_weights / exp_weights.sum()
                    exp_weights = torch.from_numpy(exp_weights).cuda().unsqueeze(dim=1)
                    raw_action = (actions_for_curr_step * exp_weights).sum(dim=0, keepdim=True)
                else:
                    raw_action = all_actions[:, t % query_frequency]

                # Post-process: denormalize, take first 7D (drop padding dim)
                action = post_process(raw_action.squeeze(0).cpu().numpy())
                action = action[:7]  # 8D -> 7D OSC_POSE

                env.step(action)
                if render:
                    env.render()

                if env._check_success():
                    episode_success = True

        successes.append(episode_success)
        status = "SUCCESS" if episode_success else "FAIL"
        print(f"  [{ep+1}/{num_rollouts}] {status}  (t={t+1})")

    success_rate = sum(successes) / len(successes)
    print(f"\n{'='*40}")
    print(f"Task: {task}")
    print(f"Checkpoint: {ckpt_dir}/{ckpt_name}")
    print(f"Mode: {'temporal_agg' if temporal_agg else 'chunk'}")
    print(f"Success: {sum(successes)}/{num_rollouts} ({success_rate*100:.1f}%)")
    print(f"{'='*40}")

    env.close()
    return success_rate


if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument("--task", type=str, required=True,
                        choices=["precision_grasp", "peg_insertion", "gentle_stack"])
    parser.add_argument("--ckpt_dir", type=str, default=None,
                        help="checkpoint dir (default: policy/EVA/checkpoints/{task}_eva)")
    parser.add_argument("--ckpt", type=str, default="policy_best.ckpt")
    parser.add_argument("--num_rollouts", type=int, default=5)
    parser.add_argument("--temporal_agg", action="store_true")
    parser.add_argument("--render", action="store_true")
    parser.add_argument("--cameras", nargs="+", default=["agentview", "eye_in_hand"])
    args = parser.parse_args()

    if args.ckpt_dir is None:
        args.ckpt_dir = os.path.join(
            os.path.dirname(__file__), "..", "policy", "EVA", "checkpoints", f"{args.task}_eva"
        )

    run_eval(
        task=args.task,
        ckpt_dir=args.ckpt_dir,
        ckpt_name=args.ckpt,
        num_rollouts=args.num_rollouts,
        temporal_agg=args.temporal_agg,
        camera_names=args.cameras,
        render=args.render,
    )