Our work “Leveraging Synthetic Data to Learn Video Stabilization Under Adverse Conditions” was presented at the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2024). Our paper presents a new technique that is trained on synthetic data in order to perform video stabilisation (on real videos). In particular, our approach also performs well under adverse weather conditions, since it does not rely on the usual feature extraction techniques. The paper is available here. Our code and datasets are also available.

Our work “Information-guided Planning: An Online Approach for Partially Observable Problems” was presented at the 37th Conference on Neural Information Processing Systems (NeurIPS 2023). The work integrates entropy into the decision-making process of the Monte Carlo simulations of an on-line planner, improving the agent’s performance, especially in scenarios with sparse rewards. The paper is freely available. Our source code is also available in the paper’s GitHub.

We presented in the Brazilian Symposium On Games and Digital Entertainment our work “Scale-Invariant Reinforcement Learning in Real-Time Strategy Games”. We integrate Spatial Pyramid Pooling (SPP) with Deep Reinforcement Learning, in order to allow a trained agent to play in maps of different dimensions in Real Time Strategy Games. The paper is freely available, including our source code.

We presented at the European Conference on Artificial Intelligence (ECAI) the work “Robust Federated Learning Method against Data and Model Poisoning Attacks with Heterogeneous Data Distribution”. The work introduces a novel technique for defending against data and model poisoning attacks in federated learning, even when there is high data heterogeneity. The paper is available for free here, and the source code is available on GitHub.

Our work “Certified Policy Smoothing for Cooperative Multi-Agent Reinforcement Learning” was presented at AAAI 2023. The paper presents a new approach to certify the robustness of multi-agent cooperative reinforcement learning systems, using policy smoothing techniques. The paper is freely available here. The source code is also available on GitHub.

We will soon present at BMVC 2022 our work “Semantic Segmentation under Adverse Conditions: A Weather and Nighttime-aware Synthetic Data-based Approach”. This work presents a novel algorithm for harnessing synthetic data in order to improve semantic segmentation in real images.

Our work is freely available here. Our source code, simulator, and new dataset can also be obtained freely at https://github.com/lsmcolab/Semantic-Segmentation-under-Adverse-Conditions.

Our work “Congestion control algorithms for robotic swarms with a common target based on the throughput of the target area” is already available on-line in the journal Robotics and Autonomous Systems. It presents novel algorithms for handling hard congestion situations when thousands of robots move towards a common target. This work is inspired by our previous theoretical work, but it is now focused on presenting algorithms for real robots.

Our author version is freely available at http://www.lancaster.ac.uk/staff/sorianom/Algorithms2023.pdf. The publisher final version is at https://www.sciencedirect.com/science/article/pii/S0921889022001737.

You can also find our source code freely available at https://github.com/yuri-tavares/swarm-common-target-area-congestion.

Our work “3DVerifier: Efficient Robustness Verification for 3D Point Cloud Models” was accepted in the journal Machine Learning. The paper is freely available at https://link.springer.com/article/10.1007/s10994-022-06235-3. It presents a novel method to verify the robustness against attacks of machine learning models aimed at classifying 3D Point Cloud data.

Our tool, 3DVerifier is also released, including its source code. You can find it at https://github.com/TrustAI/3DVerifier.

Our work “On-line estimators for ad-hoc task execution: learning types and parameters of teammates for effective teamwork” was recently published at the Journal Autonomous Agents and Multi-Agent Systems (JAAMAS). It presents OEATE, a novel algorithm for online estimation of teammates’ type and parameters in decentralised task execution.

The paper is freely available at https://link.springer.com/article/10.1007/s10458-022-09571-9. The source code, built using our AdLeap-MAS framework, is available on Github.

We recently published an extensive theoretical study on the common target problem for swarm robotics. Our work, “On the Throughput of the Common Target Area for Robotic Swarm Strategies”, was published at Mathematica, and is freely available here.

The source code of our experimented strategies are available at https://github.com/yuri-tavares/swarm-strategies.