Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
CAML: Collaborative Auxiliary Modality Learning for Multi-Agent Systems
Authors: Rui Liu, Yu Shen, Peng Gao, Pratap Tokekar, Ming C. Lin
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results in collaborative decision-making for CAV in accident-prone scenarios demonstrate that CAML achieves up to a 58.1% improvement in accident detection. Additionally, we validate CAML on real-world aerial-ground robot data for collaborative semantic segmentation, achieving up to a 10.6% improvement in mIoU. |
| Researcher Affiliation | Collaboration | Rui Liu1, Yu Shen2, Peng Gao3, Pratap Tokekar1, Ming Lin1 1University of Maryland, College Park 2Adobe Research 3North Carolina State University |
| Pseudocode | No | The paper describes its methodology in narrative text and provides an overview figure (Figure 2), but it does not contain any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper states that "Implementations are detailed in Section 4 Experiments and corresponding appendices." but does not provide a direct link to a code repository or an explicit statement that the source code is being released. |
| Open Datasets | Yes | The dataset is generated using the AUTOCASTSIM benchmark [38]... We use the dataset CoPeD [50]... [38] Hang Qiu, Pohan Huang, Namo Asavisanu, Xiaochen Liu, Konstantinos Psounis, and Ramesh Govindan. Autocast: Scalable infrastructure-less cooperative perception for distributed collaborative driving. arXiv preprint arXiv:2112.14947, 2021. [50] Yang Zhou, Long Quang, Carlos Nieto-Granda, and Giuseppe Loianno. CoPed-advancing multi-robot collaborative perception: A comprehensive dataset in real-world environments. IEEE Robotics and Automation Letters, 2024. |
| Dataset Splits | Yes | Following the setup established by prior works COOPERNAUT [5] and STGN [8], we collect 24 data trails for each scenario, using 12 trails for training and the remaining 12 for testing. |
| Hardware Specification | Yes | The model is trained on an Nvidia RTX 3090 GPU with AMD Ryzen 9 5900 CPU and 32 GB RAM for 200 epochs. |
| Software Dependencies | No | The paper mentions using the Adam optimizer and a Cosine Annealing Scheduler but does not specify versions for any programming languages, libraries, or frameworks (e.g., Python, PyTorch, TensorFlow, CUDA). |
| Experiment Setup | Yes | For the experiments, we employ a batch size of 32 and the Adam optimizer [18] with an initial learning rate of 1e-3, and a Cosine Annealing Scheduler [33] to adjust the learning rate over time. The model is trained on an Nvidia RTX 3090 GPU with AMD Ryzen 9 5900 CPU and 32 GB RAM for 200 epochs. For processing RGB data, we first resize the image to 224x224 and use ResNet-18 [14] as the encoder to extract a feature map for each vehicle... After the self-attention, we apply three convolution layers with each followed by a ReLU activation. Finally, we obtain a 256-d feature representation after passing through a fully connected layer... For processing the LiDAR data, we use the Point Transformer as the encoder for each vehicle and utilize the COOPERNAUT [5] model to aggregate LiDAR feature embeddings... Then we concatenate the final RGB and LiDAR embeddings for the ego vehicle’s decision-making, with a three-layer MLP as the prediction module to output the action. ... The soft targets are generated by applying a temperature scaling to the logits... t = 4.0 is the temperature... The overall loss for the student model is a weighted sum of the distillation loss and the binary cross-entropy loss: LS = (1 - α)LBCE(y, S) + αt2LKD(S, T ) with α = 0.9 as the weight. |