SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2603.07901 · 2026

NaviDriveVLM: Decoupling High-Level Reasoning and Motion Planning for Autonomous Driving

Gaurav Pandey, Pardis Taghavi, Reza Langari, Ximeng Tao, Dimitar Filev

arXiv · PDF · Open in the Atlas

Code that ran

2 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 1 failure of our instrument; 3 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2603.07901.json · badge: /api/ran/2603.07901.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
TAMU-CVRL/NaviDrive community
Papers with Code archive
2 ran · 3 no recorded run
FunctionStatusWhere it lives
get_vlm_inputs Ran
ran; our contract draft was wrong, not the code
TAMU-CVRL/NaviDrive/models/navidriveVLM.py
code served (permissive licence) · get_code("3a88973b89b8573c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ProcessorOutput. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
from PIL import Image
import numpy as np

torch.manual_seed(0)

class MockProcessor:
    """Mock VLM processor that implements the minimal interface needed."""
    
    def apply_chat_template(self, messages, tokenize=False, add_generation_prompt=False):
        """Returns a formatted text string from messages."""
        return "formatted_prompt_text"
    
    def __call__(self, text=None, images=None, return_tensors=None, padding=False):
        """Mock processor call that returns a dict with required tensor keys."""
        batch_size = len(text) if text else 1
        # Return a mock output with input_ids and attention_mask
        output = {
            'input_ids': torch.randint(0, 1000, (batch_size, 8)),
            'attention_mask': torch.ones((batch_size, 8)),
            'pixel_values': torch.randn((len(images), 3, 4, 4)) if images else torch.randn((1, 3, 4, 4))
        }
        # Add a to() method to the output dict wrapper
        class ProcessorOutput(dict):
            def to(self, device):
                result = ProcessorOutput()
                for k, v in self.items():
                    if isinstance(v, torch.Tensor):
                        result[k] = v.to(device)
                    else:
                        result[k] = v
                return result
        
        return ProcessorOutput(output)

# Create mock images as PIL Images
image1 = Image.new('RGB', (4, 4), color='red')
image2 = Image.new('RGB', (4, 4), color='blue')

ARGS = {
    'processor': MockProcessor(),
    'system_prompt': 'You are a helpful assistant.',
    'user_prompt': 'What do you see in these images?',
    'images': [image1, image2],
    'target_device': 'cpu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ProcessorOutput — a ProcessorOutput output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ProcessorOutput output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
parse_string Ran
ran, honoured the contract we drafted
TAMU-CVRL/NaviDrive/models/navidriveVLM.py
code served (permissive licence) · get_code("cecf40bae3294eed")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned ndarray, ndim 2, dtype family float, shape (2, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import re
import numpy as np

def parse_string(vlm_str):
    pattern = '\\(([-+]?\\d*\\.?\\d+)\\s*,\\s*([-+]?\\d*\\.?\\d+)\\s*,\\s*([-+]?\\d*\\.?\\d+)\\)'
    matches = re.findall(pattern, vlm_str)
    if not matches:
        pattern_2d = '\\(([-+]?\\d*\\.?\\d+)\\s*,\\s*([-+]?\\d*\\.?\\d+)\\)'
        matches = re.findall(pattern_2d, vlm_str)
    if not matches:
        return np.array([[]])
    results = [[float(x) for x in m] for m in matches]
    results = np.atleast_2d(np.array(results))
    return results

# Test cases with 3D coordinates
vlm_str_3d = "Point at (1.5, 2.3, -3.7) and another at (-0.5, 1.0, 2.5)"

# Test cases with 2D coordinates
vlm_str_2d = "Location (4.2, -1.8) and (0.0, 3.14)"

# Test case with no matches
vlm_str_empty = "No coordinates here"

ARGS = {
    "vlm_str": vlm_str_3d
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (2, 3) — shape (2, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
Driver Not yet run
unverified (harvested, no recorded run)
TAMU-CVRL/NaviDrive/models/navidriveVLM.py
code served (permissive licence) · get_code("e5ed40e5f879edd6")
How it ran

no invocation recorded.

NaviDriveVLM Not yet run
unverified (harvested, no recorded run)
TAMU-CVRL/NaviDrive/models/navidriveVLM.py
code served (permissive licence) · get_code("701225b930682344")
How it ran

no invocation recorded.

Navigator Not yet run
unverified (harvested, no recorded run)
TAMU-CVRL/NaviDrive/models/navidriveVLM.py
code served (permissive licence) · get_code("97624b8e85c128e1")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 5 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Vision-language models (VLMs) have emerged as a promising direction for end-to-end autonomous driving (AD) by jointly modeling visual observations, driving context, and language-based reasoning. However, existing VLM-based systems face a trade-off between high-level reasoning and motion planning: large models offer strong semantic understanding but are costly to adapt for precise control, whereas small VLM models can be fine-tuned efficiently but often exhibit weaker reasoning. We propose NaviDriveVLM, a decoupled framework that separates reasoning from action generation using a large-scale Navigator and a lightweight trainable Driver. This design preserves reasoning ability, reduces training cost, and provides an explicit interpretable intermediate representation for downstream planning. Experiments on the nuScenes benchmark show that NaviDriveVLM outperforms large VLM baselines in end-to-end motion planning. Codes are available at: https://github.com/TAMU-CVRL/NaviDrive.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2603.07901")
get_code_for_paper("2603.07901")
have("2603.07901")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2603.07901.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2603.07901.svg)](https://syntology.ai/paper/2603.07901)

Connect an agent — have() is free.