@misc{kohlbrenner2026egocentric,title={Egocentric Tactile and Proximity Sensors as Inductive Bias for Humanoid Collision Avoidance},author={Kohlbrenner, C. and Pudasaini, Niraj and Xie, W. and Roncone, Alessandro and Correll, Nikolaus},year={2026},month=may,}
FAME: Force-Adaptive RL for Expanding the Manipulation Envelope of a Full-Scale Humanoid
Niraj Pudasaini, Y. Zhang, J. Lavering, and 2 more authors
@misc{pudasaini2026fame,title={FAME: Force-Adaptive RL for Expanding the Manipulation Envelope of a Full-Scale Humanoid},author={Pudasaini, Niraj and Zhang, Y. and Lavering, J. and Roncone, Alessandro and Correll, Nikolaus},year={2026},url={https://fame10.github.io/Fame/},note={In Preparation}}
2025
H2-COMPACT: Human-Humanoid Co-Manipulation via Adaptive Contact Trajectory Policies
Geeta Chandra Raju Bethala, Hao Huang, Niraj Pudasaini, and 5 more authors
We present a hierarchical policy-learning framework that enables a legged humanoid to cooperatively carry extended loads with a human partner using only haptic cues for intent inference. At the upper tier, a lightweight behaviorcloning network consumes six-axis force/torque streams from dual wrist-mounted sensors and outputs whole-body planar velocity commands that capture the leader’s applied forces. At the lower tier, a deep-reinforcement-learning policy—trained under randomized payloads (0–3 kg) and friction conditions in Isaac Gym and validated in MuJoCo and on a real Unitree G1—maps these high-level twists to stable, under-load joint trajectories. By decoupling intent interpretation (force → velocity) from legged locomotion (velocity → joints), our method combines intuitive responsiveness to human inputs with robust, load-adaptive walking. We collect training data without motion-capture or markers—only synchronized RGB video and F/T readings—employing SAM2 and WHAM to extract 3D human pose and velocity. In real-world trials, our humanoid achieves cooperative carry-and-move performance (completion time, trajectory deviation, velocity synchrony, and follower-force) on par with a blindfolded human-follower baseline. This work is the first to demonstrate learned haptic guidance fused with full-body legged control for fluid human–humanoid co–manipulation. Code and videos are available on https://h2compact.github.io/h2compact/.
@misc{bethala2025h2compacthumanhumanoidcomanipulationadaptive,title={H2-COMPACT: Human-Humanoid Co-Manipulation via Adaptive Contact Trajectory Policies},author={Bethala, Geeta Chandra Raju and Huang, Hao and Pudasaini, Niraj and Ali, Abdullah Mohamed and Yuan, Shuaihang and Wen, Congcong and Tzes, Anthony and Fang, Yi},year={2025},eprint={2505.17627},archiveprefix={arXiv},primaryclass={cs.RO},url={https://www.arxiv.org/abs/2505.17627},publisher={IEEE-RAS International Conference on Humanoid Robots (Humanoids)},}
LAV-ACT: Language-Augmented Visual ACT for Bimanual Manipulation
Dhurba Tripathi, Chunting Liu, Niraj Pudasaini, and 3 more authors
In Proceedings of the IEEE International Conference on Autonomous Robot Systems and Competitions (ICARA), Feb 2025
@inproceedings{tripathi2025lavact,title={LAV-ACT: Language-Augmented Visual ACT for Bimanual Manipulation},author={Tripathi, Dhurba and Liu, Chunting and Pudasaini, Niraj and Hao, Yu and Tzes, Anthony and Fang, Yi},booktitle={Proceedings of the IEEE International Conference on Autonomous Robot Systems and Competitions (ICARA)},year={2025},month=feb,url={https://ieeexplore.ieee.org/document/10977578},publisher={IEEE International Conference on Autonomous Robot Systems and Competitions (ICARA)},note={Equal contribution: D. Tripathi, C. Liu}}
ROVER: Autonomous Open-Vocabulary Object Searching in Unexplored Environments Using VLM-Driven Scene Understanding
Abdul Basit, Niraj Pudasaini, and Muhammad Shafique
In Proceedings of the IEEE International Joint Conference on Neural Networks (IJCNN), Jul 2025
@inproceedings{basit2025rover,title={ROVER: Autonomous Open-Vocabulary Object Searching in Unexplored Environments Using VLM-Driven Scene Understanding},author={Basit, Abdul and Pudasaini, Niraj and Shafique, Muhammad},booktitle={Proceedings of the IEEE International Joint Conference on Neural Networks (IJCNN)},year={2025},month=jul,url={https://ieeexplore.ieee.org/document/11228732},publisher={IEEE International Joint Conference on Neural Networks (IJCNN)},}
Embodied Chain of Action Reasoning with Multi-Modal Foundation Model for Humanoid Loco-manipulation
Yu Hao*, Geeta Chandra Raju Bethala*, Niraj Pudasaini*, and 6 more authors
Enabling humanoid robots to autonomously perform loco-manipulation tasks in complex, unstructured environments poses significant challenges. This entails equipping robots with the capability to plan actions over extended horizons while leveraging multi-modality to bridge gaps between high-level planning and actual task execution. Recent advancements in multi-modal foundation models have showcased substantial potential in enhancing planning and reasoning abilities, particularly in the comprehension and processing of semantic information for robotic control tasks. In this paper, we introduce a novel framework based on foundation models that applies the embodied chain of action reasoning methodology to autonomously plan actions from textual instructions for humanoid loco-manipulation. Our method integrates humanoid-specific chain of thought methodology, including detailed affordance and body movement analysis, which provides a breakdown of the task into a sequence of locomotion and manipulation actions. Moreover, we incorporate spatial reasoning based on the observation and target object properties to effectively navigate where target position may be unseen or occluded. Through rigorous experimental setups on object rearrangement, manipulations and loco-manipulation tasks on a real-world environment, we evaluate our method’s efficacy on the decoupled upper and lower body control and demonstrate the effectiveness of the chain of robotic action reasoning strategies in comprehending human instructions.
@misc{hao2025embodiedchainactionreasoning,title={Embodied Chain of Action Reasoning with Multi-Modal Foundation Model for Humanoid Loco-manipulation},author={Hao, Yu and Bethala, Geeta Chandra Raju and Pudasaini, Niraj and Huang, Hao and Yuan, Shuaihang and Wen, Congcong and Huang, Baoru and Nguyen, Anh and Fang, Yi},year={2025},eprint={2504.09532},archiveprefix={arXiv},primaryclass={cs.RO},url={https://arxiv.org/abs/2504.09532},}
2024
SPAQ-DL-SLAM: Towards Optimizing Deep Learning-based SLAM for Resource-Constrained Embedded Platforms
Niraj Pudasaini, Muhammad Abdullah Hanif, and Muhammad Shafique
Optimizing Deep Learning-based Simultaneous Localization and Mapping (DL-SLAM) algorithms is essential for efficient implementation on resource-constrained embedded platforms, enabling real-time on-board computation in autonomous mobile robots. This paper presents SPAQ-DL-SLAM, a framework that strategically applies Structured Pruning and Quantization (SPAQ) to the architecture of one of the state-ofthe-art DL-SLAM algorithms, DROID-SLAM, for resource and energy-efficiency. Specifically, we perform structured pruning with fine-tuning based on layer-wise sensitivity analysis followed by 8-bit post-training static quantization (PTQ) on the deep learning modules within DROID-SLAM. Our SPAQ-DROIDSLAMmodel, optimized version of DROID-SLAM model using our SPAQ-DL-SLAM framework with 20% structured pruning and 8-bit PTQ, achieves an 18.9% reduction in FLOPs and a 79.8% reduction in overall model size compared to the DROID-SLAM model. Our evaluations on the TUM-RGBD benchmark shows that SPAQ-DROID-SLAM model surpasses the DROID-SLAM model by an average of 10.5% on absolute trajectory error (ATE) metric. Additionally, our results on the ETH3D SLAM training benchmark demonstrate enhanced generalization capabilities of the SPAQ-DROID-SLAM model, seen by a higher Area Under the Curve (AUC) score and success in 2 additional data sequences compared to the DROIDSLAM model. Despite these improvements, the model exhibits performance variance on the distinct Vicon Room sequences from the EuRoC dataset, which are captured at high angular velocities. This varying performance at some distinct scenarios suggests that designing DL-SLAM algorithms taking operating environments and tasks in consideration can achieve optimal performance and resource efficiency for deployment in resource-constrained embedded platforms.
@misc{pudasaini2024spaqdlslamoptimizingdeeplearningbased,title={SPAQ-DL-SLAM: Towards Optimizing Deep Learning-based SLAM for Resource-Constrained Embedded Platforms},author={Pudasaini, Niraj and Hanif, Muhammad Abdullah and Shafique, Muhammad},year={2024},eprint={2409.14515},archiveprefix={arXiv},primaryclass={cs.RO},url={https://arxiv.org/abs/2409.14515},publisher={International Conference on Control, Automation, Robotics and Vision - ICARCV },}
Immersive Social Interaction with VR and LLM-Assisted Humanoids
Niraj Pudasaini, Geeta Chandra Raju Bethala, Nikolaos Giakoumidis, and 2 more authors
In IEEE-RAS International Conference on Humanoid Robots (Humanoids) Workshop, Nov 2024
Spotlight Paper. Equal contribution: N. Pudasaini, G. Bethala
@inproceedings{pudasaini2024vr,title={Immersive Social Interaction with VR and LLM-Assisted Humanoids},author={Pudasaini, Niraj and Bethala, Geeta Chandra Raju and Giakoumidis, Nikolaos and Tzes, Anthony and Fang, Yi},booktitle={IEEE-RAS International Conference on Humanoid Robots (Humanoids) Workshop},year={2024},month=nov,url={https://www.researchgate.net/publication/394225021},publisher={IEEE-RAS International Conference on Humanoid Robots (Humanoids) Workshop},note={Spotlight Paper. Equal contribution: N. Pudasaini, G. Bethala}}
