We are delighted to share that two papers from RoboLab have been accepted for presentation at the 6th Muslims in Machine Learning (MusIML) Workshop, held in conjunction with the 43rd International Conference on Machine Learning (ICML) 2026.
Hafiq’s paper was on “Crowd Navigation for Mobile Robots with Focused Risk Perception”.
Current state-of-the-art crowd navigation approaches are mainly deep reinforcement learning (DRL)-based but often struggle to generalize to unseen crowd behaviors and to scale efficiently with crowd density. We propose a DRL method that incorporates risk perception directly into the observation space: collision probability is used to identify the K most hazardous obstacles, whose relative positions and velocities are exposed to the policy, and local waypoints are added to the reward to densify the learning signal. Trained once in the Gazebo simulator with a non-cooperative randomized crowd, our model is evaluated on four crowd-behavior scenarios against a 2D-laser DRL baseline, a human-aware social planner, and two classical planners. Our approach achieves higher success rates and social safety across all settings, generalizes to unseen crowd behaviors without fine-tuning, and is further validated in real-world tests.
Nazrul’s paper was on “Evaluating Visual-to-Echo Distillation for Binaural Depth Prediction beyond Simulations”.
Cross-modal vision-to-echo distillation has largely been studied in simulation, with limited validation on real-world data. Using real binaural recordings from BatVision (BV1), we evaluate V2E-CCL and demonstrate a 15.8% reduction in relative error (REL) over the strongest audio-only baseline, while an ablation shows that the gain stems from the distillation strategy rather than the backbone architecture.
At MusIML@ICML 2026, we presented the posters of Hafiq and Nazrul. We are especially honored that Hafiq was selected to deliver an oral presentation at the workshop. Unfortunately, Nazrul was unable to attend the conference as he was participating in the International Computer Vision Summer School (ICVSS) 2026, which took place during the same period. Instead, Azryl join us on this trip to experience the big event.
We are grateful to the Muslims in Machine Learning (MusIML) community for generously supporting our registration. It was a privilege to meet, connect, and exchange ideas with the inspiring researchers and students within the MusIML community. Your support provided us with the valuable opportunity to participate in one of the world’s leading conferences in machine learning. It was a pleasure meeting and connecting with the wonderful community at MusIML@ICML. Dr. Ajaz also has a paper accepted at the Workshop on Foundations of Deep Generative Models: Understanding Memorization, Generalization, and Reasoning.
ICML 2026 was a remarkable gathering of the global machine learning community, with more than 20k in-person attendees. The conference received over 24k paper submissions from more than 76k authors, with approximately 6.5k papers accepted (an acceptance rate of 26.6%). While we did not have a paper at the main ICML conference this year, attending ICML was an invaluable experience that allowed us to learn from cutting-edge research, engage with leading researchers, and gain inspiration for our future work.
