File size: 8,310 Bytes
f9b94bf 80d32d0 f9b94bf 80d32d0 f9b94bf 80d32d0 f9b94bf | 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 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 | ---
license: apache-2.0
library_name: transformers
pipeline_tag: robotics
tags:
- vision-language-action
- robotics
- embodied-ai
- minicpm
- visual-tracking
---
<h1 align="center">
MiniCPM-RobotTrack
</h1>
<p align="center">
<strong>A Compact Vision-Language-Action Policy for Embodied target Tracking</strong>
</p>
<p align="center">
<span style="display: inline-flex; align-items: center; margin-right: 2px;">
<img src="assets/github.png" alt="github" width="15" height="15" style="margin-right: 4px;">
<a href="https://github.com/OpenBMB/MiniCPM-Robot" target="_blank"> Github</a> |
</span>
<span style="display: inline-flex; align-items: center; margin-right: 2px;">
<img src="assets/x.jpg" alt="X" width="15" height="15" style="margin-right: 4px;">
<a href="https://huggingface.co/openbmb/MiniCPM-RobotTrack/resolve/main/assets/x.png" target="_blank"> X</a> |
</span>
<span style="display: inline-flex; align-items: center; margin-right: 2px;">
<img src="assets/discord.png" alt="Discord" width="15" height="15" style="margin-right: 4px;">
<a href="https://huggingface.co/openbmb/MiniCPM-RobotTrack/resolve/main/assets/discord.jpg" target="_blank"> Discord</a> |
</span>
</p>
<strong>MiniCPM-RobotTrack</strong> is a compact vision-language-action model built on MiniCPM4-0.5B for target tracking with the following highlights:
<ul>
<li><b>Language-conditioned target tracking:</b> The model combines natural-language instructions with fused DINOv3 and SigLIP visual features, then directly predicts eight future <code>[x, y, yaw]</code> waypoints for embodied person following.</li>
<li><b>Quality-driven Self-evolving Data:</b> Automated checks and manual review remove abnormal trajectories, incorrect actions, and invalid interactions. DAgger-style model-environment interaction continually adds difficult samples involving target crossings, rapid turns, short occlusions, and multi-person intersections.</li>
<li><b>Efficient On-device Inference:</b> Joint optimization of visual capture, input encoding, model inference, action generation, command transmission, and execution delivers a stable <b>5+ FPS</b> with approximately <b>180 ms</b> end-to-end latency on the Unitree Go2's native onboard compute.</li>
</ul>
<table align="center">
<tr>
<td align="center" width="33%">
<img src="assets/track1.gif" width="100%" alt="Outdoor obstacle-aware person-tracking demo" />
</td>
<td align="center" width="33%">
<img src="assets/track2.gif" width="100%" alt="Elevator person-tracking demo" />
</td>
<td align="center" width="33%">
<img src="assets/track3.gif" width="100%" alt="Underground parking person-tracking demo" />
</td>
</tr>
<tr>
<td align="center"><b>Outdoor Obstacle-aware Tracking</b></td>
<td align="center"><b>Elevator Tracking</b></td>
<td align="center"><b>Underground Parking Tracking</b></td>
</tr>
</table>
## Benchmark Results
<p align="center">
<img src="assets/metric.png" width="700" alt="MiniCPM-RobotTrack benchmark results" />
</p>
## Inference Example
Please ensure `transformers>=4.56,<5`.
<pre><code class="language-python">from __future__ import annotations
from pathlib import Path
import torch
from transformers import AutoModel, AutoTokenizer
VISION_FEATURE_DIM = 1536
HISTORY_FRAMES = 31
COARSE_TOKENS_PER_FRAME = 4
FINE_TOKENS_CURRENT_FRAME = 64
class MiniCPMRobotTrackInference:
"""Tokenizer and model wrapper for MiniCPM-RobotTrack inference."""
def __init__(
self,
checkpoint_path: str | Path = "openbmb/MiniCPM-RobotTrack",
device: str | torch.device | None = None,
):
if device is None:
device = "cuda" if torch.cuda.is_available() else "cpu"
self.device = torch.device(device)
checkpoint = str(checkpoint_path)
self.tokenizer = AutoTokenizer.from_pretrained(checkpoint)
if self.tokenizer.pad_token_id is None:
self.tokenizer.pad_token = self.tokenizer.eos_token
self.model = AutoModel.from_pretrained(
checkpoint,
trust_remote_code=True,
)
self.model.to(self.device).eval()
@staticmethod
def _prepare_visual_tokens(
tokens: torch.Tensor,
time_indices: torch.Tensor,
name: str,
) -> tuple[torch.Tensor, torch.Tensor]:
tokens = torch.as_tensor(tokens, dtype=torch.float32)
time_indices = torch.as_tensor(time_indices, dtype=torch.long)
if tokens.ndim == 2:
tokens = tokens.unsqueeze(0)
if time_indices.ndim == 1:
time_indices = time_indices.unsqueeze(0)
if tokens.ndim != 3 or tokens.shape[-1] != VISION_FEATURE_DIM:
raise ValueError(
f"{name}_tokens must have shape [B, N, {VISION_FEATURE_DIM}]"
)
if time_indices.shape != tokens.shape[:2]:
raise ValueError(
f"{name}_time_indices must match the first two dimensions of "
f"{name}_tokens"
)
return tokens, time_indices
@torch.inference_mode()
def predict(
self,
instruction: str,
coarse_tokens: torch.Tensor,
coarse_time_indices: torch.Tensor,
fine_tokens: torch.Tensor,
fine_time_indices: torch.Tensor,
) -> torch.Tensor:
"""Return eight ``[x, y, yaw]`` waypoints for each batch item."""
coarse_tokens, coarse_time_indices = self._prepare_visual_tokens(
coarse_tokens,
coarse_time_indices,
"coarse",
)
fine_tokens, fine_time_indices = self._prepare_visual_tokens(
fine_tokens,
fine_time_indices,
"fine",
)
batch_size = coarse_tokens.shape[0]
if fine_tokens.shape[0] != batch_size:
raise ValueError("coarse and fine feature batch sizes must match")
text = self.tokenizer(
[instruction] * batch_size,
return_tensors="pt",
padding=True,
truncation=True,
max_length=self.model.config.max_text_tokens,
)
outputs = self.model(
input_ids=text.input_ids.to(self.device),
attention_mask=text.attention_mask.to(self.device),
coarse_tokens=coarse_tokens.to(self.device),
coarse_time_indices=coarse_time_indices.to(self.device),
fine_tokens=fine_tokens.to(self.device),
fine_time_indices=fine_time_indices.to(self.device),
)
return outputs.trajectories.float().cpu()
if __name__ == "__main__":
infer_runner = MiniCPMRobotTrackInference()
# Replace these placeholders with fused DINOv3 + SigLIP features produced
# by the project preprocessing pipeline.
coarse_tokens = torch.zeros(
HISTORY_FRAMES * COARSE_TOKENS_PER_FRAME,
VISION_FEATURE_DIM,
)
coarse_time_indices = torch.arange(HISTORY_FRAMES).repeat_interleave(
COARSE_TOKENS_PER_FRAME
)
fine_tokens = torch.zeros(
FINE_TOKENS_CURRENT_FRAME,
VISION_FEATURE_DIM,
)
fine_time_indices = torch.full(
(FINE_TOKENS_CURRENT_FRAME,),
HISTORY_FRAMES,
dtype=torch.long,
)
trajectory = infer_runner.predict(
instruction="Follow the person in the red shirt.",
coarse_tokens=coarse_tokens,
coarse_time_indices=coarse_time_indices,
fine_tokens=fine_tokens,
fine_time_indices=fine_time_indices,
)
print(trajectory) # [1, 8, 3]
</code></pre>
## Acknowledgement
<p>
This project builds on and references
<a href="https://github.com/OpenBMB/MiniCPM">MiniCPM</a>,
<a href="https://github.com/facebookresearch/dinov3">DINOv3</a>,
<a href="https://huggingface.co/google/siglip-so400m-patch14-384">SigLIP</a>,
<a href="https://github.com/facebookresearch/habitat-lab">Habitat-Lab</a>,
<a href="https://github.com/facebookresearch/habitat-sim">Habitat-Sim</a>,
EVT-Bench,
and
<a href="https://github.com/wsakobe/TrackVLA">TrackVLA</a>.
We thank the authors for their open-source contributions.
</p>
## License
Model weights and code are open-sourced under the [Apache-2.0](./LICENSE) license.
|