IPAB Workshop-14-3-24
Speaker: Zhipeng Du
Title: Boosting Object Detection with Zero-Shot Day-Night Domain-Adaptation
Abstract: Detecting objects in low-light scenarios presents a persistent challenge, as detectors trained on well-lit data exhibit significant performance degradation on low-light data due to the low visibility. Previous methods mitigate this issue by investigating image enhancement or object detection techniques using low-light image datasets. However, the progress is impeded by the inherent difficulties associated with collecting and annotating low-light images. To address this challenge, we propose to boost low-light object detection with zero-shot day-night domain adaptation, which aims to generalize a detector from well-lit scenarios to low-light ones without requiring real low-light data. We first design a reflectance representation learning module to learn Retinex-based illumination invariance in images with a carefully designed illumination invariance reinforcement strategy. Next, an interchange-redecomposition-coherence procedure is introduced to improve over the vanilla Retinex image decomposition process by performing two sequential image decompositions and introducing a redecomposition cohering loss. Extensive experiments on ExDark, DARK FACE and CODaN datasets show strong low-light generalizability of our method.
Speaker: Aditya Kamireddypalli
Title: Hybrid state estimation for robot assembly tasks
Abstract: Hybrid State Estimation is a crucial step in contact-rich robotic settings, particularly when synthesizing reactive behaviour in settings that involve making and breaking contact. A simple example task is that of part mating, wherein a part needs to be inserted into another with tight tolerances. Using a combination of visual and proprioceptive cues, estimation of whether the insertion has been completed or if the part is stuck in an intermediate state is nontrivial, and serves as an example of situations where hybrid state estimation is needed. We address this problem using a factor graph formulation, combined with articulation models. This talk represents a status report of current work in progress.
Speaker: Namiko Saito
Title: What Does It Take to Realize Cooking Robots?: From Scientific Research to Real-World Application for Cooking Robotics
Abstract: In this presentation, I summarize and categorize the essential elements required for cooking robots, covering hardware, task planning, perception, motion planning and control, and application. I'll showcase some of my research work on perception, motion planning, and task planning, illustrating advancements and challenges in enabling robots to perform culinary tasks autonomously.
Furthermore, I'll discuss the significance of cooking robot research, addressing scientific contributions, social impact, and real-world challenges.