29 June 2022 in IEEE Robotics and Automation Letters
We propose GA-Nav, a novel group-wise attention mechanism to identify safe and navigable regions in unstructured environments from RGB images. Our group-wise attention method extracts multi-scale features from each type of terrain independently and classifies terrains based on their navigability levels using coarse-grained semantic segmentation. Our novel loss can be embedded within any backbone network to explicitly focus on the different groups’ features, at a low spatial resolution. Our design leads to efficient inference while maintaining a high level of accuracy compared to existing SOTA methods. Our extensive evaluations on the RUGD and RELLIS-3D datasets shows that GA-Nav achieves the state-of-the-art performance on RUGD and RELLIS-3D datasets. We interface GA-Nav with a deep reinforcement learning-based navigation algorithm and highlight its benefits in terms of navigation in real-world unstructured terrains. We integrate our GA-Nav-based navigation algorithm with ClearPath Jackal and Husky robots, and observe an improvement in terms of navigation success rate and better trajectory selections.
7 October 2026 in Discover Artificial Intelligence
Specifying nuanced and compelling camera motion remains a significant hurdle for non-expert creators using generative tools, creating an “expressive gap” where generic text prompts fail to capture cinematic vision. This barrier limits individual creativity and restricts the accessibility of cinematic production for small-scale industries and educational content creators. To address this, we present a zero-shot diffusion-based framework for personalized camera motion control, enabling the transfer of cinematic movements from a single reference video onto a user-provided static image without requiring 3D data, predefined trajectories, or complex graphical interfaces. Our technical contribution involves an inference-time optimization strategy using dual Low-Rank Adaptation (LoRA) networks, with an orthogonality regularizer that encourages separation between spatial appearance and temporal motion updates, alongside a homography-based refinement strategy that provides weak geometric guidance. We evaluate our approach using a new metric, CameraScore, and two distinct user studies. A 72-participant perceptual study demonstrates that our method significantly outperforms existing baselines in motion accuracy (90.45% preference) and scene preservation (70.31% preference). Furthermore, a 12-participant task-based interaction study confirms that our workflow significantly improves usability and creative control (p < 0.001) compared to standard text- or preset-based prompts. We hope this work lays a foundation for future advancements in camera motion transfer across diverse scenes.