SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.01362 · 2023

Robot Fleet Learning via Policy Merging

arXiv · PDF · Open in the Atlas

Code that ran

8 of 12 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; 7 failures of our instrument; 4 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2310.01362.json · badge: /api/ran/2310.01362.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
liruiw/fleet-tools canonical (classifier label)
classifier basis: census
8 ran · 4 no recorded run
FunctionStatusWhere it lives
inv_lookat Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
liruiw/fleet-tools/core/utils/geometry.py
code served (permissive licence) · get_code("c58d54407128ffe2")
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 ndarray, ndim 2, dtype family float, shape (3, 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 numpy as np

# Create small valid arguments for inv_lookat function
eye = np.array([1.0, 2.0, 3.0], dtype=np.float32)
target = [0.0, 0.0, 0.0]
up = [0.0, 1.0, 0.0]

ARGS = {
    "eye": eye,
    "target": target,
    "up": up
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 (3, 3) — shape (3, 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
make_empty_dofset Ran
ran; our contract draft was wrong, not the code
liruiw/fleet-tools/core/utils/spatialstiffness_controller.py
code served (permissive licence) · get_code("092df54141d45dcb")
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 DofSetForDynamics. 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 numpy as np
import dataclasses as dc

@dc.dataclass
class DofSetForDynamics:
    """
    Like DofSet, but for dynamics where we are not constrained to nq = nv = nu.
    """
    q: np.ndarray
    v: np.ndarray
    u: np.ndarray

def make_empty_dofset(plant):
    return DofSetForDynamics(q=np.zeros(plant.num_positions(), dtype=bool), v=np.zeros(plant.num_velocities(), dtype=bool), u=np.zeros(plant.num_actuated_dofs(), dtype=bool))

# Create a minimal plant stand-in class that implements the required methods
class MockPlant:
    def __init__(self, num_positions=4, num_velocities=5, num_actuated_dofs=3):
        self._num_positions = num_positions
        self._num_velocities = num_velocities
        self._num_actuated_dofs = num_actuated_dofs
    
    def num_positions(self):
        return self._num_positions
    
    def num_velocities(self):
        return self._num_velocities
    
    def num_actuated_dofs(self):
        return self._num_actuated_dofs

# Create a mock plant with distinct dimensions
plant = MockPlant(num_positions=4, num_velocities=5, num_actuated_dofs=3)

ARGS = {
    "plant": plant,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DofSetForDynamics — a DofSetForDynamics output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DofSetForDynamics 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
normalize Ran
ran; our contract draft was wrong, not the code
liruiw/fleet-tools/core/utils/geometry.py
code served (permissive licence) · get_code("0ac6280f136b8080")
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 ndarray, ndim 2, dtype family float, shape (2, 4) 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 numpy as np

# Create a small test vector
v = np.random.randn(2, 4)

# Define ARGS dictionary with all parameters
ARGS = {
    "v": v,
    "axis": None,
    "eps": 1e-10
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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, 4) — shape (2, 4) 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
parse_buttons Ran
ran; our contract draft was wrong, not the code
liruiw/fleet-tools/core/utils/oculus_reader.py
code served (permissive licence) · get_code("b0cc76719ba6bda3")
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 dict. 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.

# Analyze the function parse_buttons
# It takes a single parameter 'text' which is a string
# The function:
# 1. Splits text by commas
# 2. Checks for 'R' and 'L' substrings and removes them
# 3. Updates a buttons dict with initial False values for certain keys
# 4. Checks if button keys are in the split text and sets them to True
# 5. Parses remaining elements as key-value pairs with space-separated floats

# Construct a valid text argument that exercises the function

text = "R,L,A,X,Y 1.5 2.3,RJ 0.5,LTr 1.0 2.0 3.0"

ARGS = {
    "text": text
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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_task Ran
ran; our contract draft was wrong, not the code
liruiw/fleet-tools/core/experiment.py
code served (permissive licence) · get_code("4a6452e0b6026f5d")
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 Mock. 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 numpy as np
from unittest.mock import Mock, MagicMock

torch.manual_seed(0)

# Create a minimal config object with the required structure
class TaskConfig:
    def __init__(self):
        self.name = "test_task"

class Config:
    def __init__(self):
        self.task = TaskConfig()

# Create mock functions for register_env and make_env
def register_env(name):
    """Mock implementation of register_env"""
    pass

def make_env(name, task_config=None):
    """Mock implementation of make_env that returns a minimal environment"""
    env = Mock()
    env.name = name
    env.task_config = task_config
    return env

# Inject the mocks into the module scope so parse_task can use them
import sys
current_module = sys.modules[__name__]
current_module.register_env = register_env
current_module.make_env = make_env

# Build the config object
cfg = Config()

# Construct ARGS
ARGS = {
    "cfg": cfg
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Mock — a Mock output has no array shape of its own; none is 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
quat_mul Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
liruiw/fleet-tools/core/utils/torch_utils.py
code served (permissive licence) · get_code("6a8cef50eba70ad0")
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 Tensor, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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

torch.manual_seed(0)

# The function quat_mul takes two quaternion tensors and multiplies them
# It expects tensors of shape (..., 4) where the last dimension is [x, y, z, w]
# We'll create two small batch tensors with shape (2, 4)

a = torch.randn(2, 4)
b = torch.randn(2, 4)

ARGS = {
    "a": a,
    "b": b,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
rotZ Ran
ran, honoured the contract we drafted
liruiw/fleet-tools/core/utils/geometry.py
code served (permissive licence) · get_code("c2983aaf729c9cd7")
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 ndarray, ndim 2, dtype family float, shape (4, 4) 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 numpy as np

ARGS = {
    "rotz": np.pi / 4
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 (4, 4) — shape (4, 4) 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
to_torch Ran
ran; our contract draft was wrong, not the code
liruiw/fleet-tools/core/utils/torch_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from gewu-lab/llm_articulated_object_manipulation, is pointer only (licence: NONE) · get_code("c473618f59887d59")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from gewu-lab/llm_articulated_object_manipulation, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

dense_sample_traj_times Not yet run
unverified (harvested, no recorded run)
liruiw/fleet-tools/core/expert/base_expert.py
code served (permissive licence) · get_code("3432463b949e1ef9")
How it ran

no invocation recorded.

get_keyboard_input Not yet run
unverified (harvested, no recorded run)
liruiw/fleet-tools/core/utils/teleop_utils.py
code served (permissive licence) · get_code("88de023c40c564f9")
How it ran

no invocation recorded.

get_mouse_input Not yet run
unverified (harvested, no recorded run)
liruiw/fleet-tools/core/utils/teleop_utils.py
code served (permissive licence) · get_code("11a995a61162212f")
How it ran

no invocation recorded.

rotation_matrix_to_axang3 Not yet run
unverified (harvested, no recorded run)
liruiw/fleet-tools/core/utils/spatialstiffness_controller.py
code served (permissive licence) · get_code("14727c675add3b59")
How it ran

no invocation recorded.

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 12 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 11 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.

FunctionAlso present inEarliest-datedOthers named
to_torch 1 other paper
1 later
2311.02847 (2023-11)
Kinematic-aware Prompting for Generalizable Articulated Object Manipulation with LLMs

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

Fleets of robots ingest massive amounts of heterogeneous streaming data silos generated by interacting with their environments, far more than what can be stored or transmitted with ease. At the same time, teams of robots should co-acquire diverse skills through their heterogeneous experiences in varied settings. How can we enable such fleet-level learning without having to transmit or centralize fleet-scale data? In this paper, we investigate policy merging (PoMe) from such distributed heterogeneous datasets as a potential solution. To efficiently merge policies in the fleet setting, we propose FLEET-MERGE, an instantiation of distributed learning that accounts for the permutation invariance that arises when parameterizing the control policies with recurrent neural networks. We show that FLEET-MERGE consolidates the behavior of policies trained on 50 tasks in the Meta-World environment, with good performance on nearly all training tasks at test time. Moreover, we introduce a novel robotic tool-use benchmark, FLEET-TOOLS, for fleet policy learning in compositional and contact-rich robot manipulation tasks, to validate the efficacy of FLEET-MERGE on the benchmark.

For agents

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

get_harvested_code_for_paper("2310.01362")
get_code_for_paper("2310.01362")
have("2310.01362")

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

curl https://syntology.ai/api/ran/2310.01362.json

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

[![Syntology run record](https://syntology.ai/api/ran/2310.01362.svg)](https://syntology.ai/paper/2310.01362)

Connect an agent — have() is free.