MSR Thesis Defense

Abstract: Designing and developing robots that can effectively navigate real-world environments poses a significant challenge. To overcome this, many robotic systems draw inspiration from the adaptive behaviors of animals, which have evolved to thrive in diverse surroundings. Amphibious animals, for instance, seamlessly transition between walking and swimming, optimizing their locomotion efficiency based on environmental cues. [...]

Abstract: Recent advancements in vision-language foundation models, exemplified by GPT4-Vision and DALL-E 3, have significantly transformed both research and practical applications, ranging from professional assistance to content creation. However, aligning them precisely with specific user goals presents a notable challenge. This thesis introduces innovative strategies for improving this alignment. I will first introduce our novel [...]

Abstract: Modern methods solve multi-object tracking from two perspectives: motion modeling and appearance matching. As a classic paradigm, motion-based tracking by Kalman filters suffers from complicated motion patterns and the problem becomes more difficult when we only have noisy bounding boxes. To improve Kalman filter-based multi-object tracking in scenarios with complex motion, occlusion, and crossover, [...]

Abstract: Modern methods solve multi-object tracking from two perspectives: motion modeling and appearance matching. As a classic paradigm, motion-based tracking by Kalman filters suffers from complicated motion patterns and the problem becomes more difficult when we only have noisy bounding boxes. To improve Kalman filter-based multi-object tracking in scenarios with complex motion, occlusion, and crossover, [...]

Abstract: Dense simultaneous localization and mapping (SLAM) is crucial for numerous robotic and augmented reality applications. However, current methods are often hampered by the non-volumetric or implicit way they represent a scene. This talk introduces SplaTAM, an approach that leverages explicit volumetric representations, i.e., 3D Gaussians, to enable high-fidelity reconstruction from a single unposed RGB-D [...]

Abstract: In recent years, researchers have extensively used non-verbal gestures, such as head and arm movements, to express the robot's intentions and capabilities to humans. Inspired by past research, we investigated how different explanation modalities can aid human understanding and perception of how robots communicate failures and provide explanations during block pick-and-place tasks. Through an in-person [...]

Abstract: Relative placement tasks are an important category of tasks in which one object needs to be placed in a desired pose relative to another object.  Previous work has shown success in learning relative placement tasks from just a small number of demonstrations, when using relational reasoning networks with geometric inductive biases. However, such methods fail [...]

Abstract: This thesis introduces a novel transfer learning framework for deep reinforcement learning. The approach automatically combines goal-conditioned policies with temporal contrastive learning to discover meaningful sub-goals. The approach involves pre-training a goal-conditioned agent, finetuning it on the target domain, and using contrastive learning to construct a planning graph that guides the agent via sub-goals. Experiments [...]

Abstract: Accurate representation in media is known to improve the well-being of the people who consume it. There is a growing concern about the increasing use of generative AI in media as the generative image models trained on large web-crawled datasets such as LAION are known to produce images with harmful stereotypes and misrepresentations of various groups, [...]

Abstract: We propose UpFusion, a system that can perform novel view synthesis and infer 3D representations for generic objects given a sparse set of reference images without corresponding pose information. Current sparse-view 3D inference methods typically rely on camera poses to geometrically aggregate information from input views, but are not robust in-the-wild when such information [...]