π Transportation - Bay Area Water Transport Scaling
This article discusses the need for streamlined regulations to facilitate the expansion of water transportation infrastructure in the San Francisco Bay Area, proposing high-speed water shuttles as a potential solution.
Key Points:
β’ Reducing bureaucratic hurdles for dock and marina construction.
β’ Simplifying dredging processes for improved waterway access.
β’ Enabling the development of high-speed water transit options.
π Resources:
β’ Sampriti Bhattacharya's Twitter β - Advocate for Bay Area water transport
Image
π€ AI - Visual Reasoning with Mirage
This article introduces Mirage, a system enabling AI to process visual information through mental imagery without actual image generation.
Key Points:
β’ Enables AI to "think in pictures" without creating visual representations.
β’ Facilitates visual reasoning capabilities in AI systems.
β’ Mimics human-like imagination for robots.
π Resources:
β’ Revanth Atmakuri's Twitter β - Shared information about Mirage
Image
π€ VFX - Nano Atmospheric Haze Plugin
This article describes Nano, a depth-aware plugin leveraging machine learning for realistic fog and light scattering effects in video footage.
Key Points:
β’ Employs ML depth estimation for physically accurate fog simulation.
β’ Optimized for log footage with visible light sources.
β’ Analyzes scene highlights to generate airlight and atmospheric scatter.
π Resources:
β’ Bilawal Sidhu's Twitter β - Shared information about Nano
Image
π€ 3D Printing - Cable Chain and Conformal Wire Guiding
This article details design improvements for a 3D-printed assembly, including the addition of a cable chain, conformal wire guiding, and a shell for easier printing.
Key Points:
β’ Improved cable management with a cable chain.
β’ Enhanced wire routing with conformal guiding.
β’ Shelled assembly for simplified 3D printing.
π Resources:
β’ Cooper Zurad's Twitter β - Shared design updates and images
Image
Image
π‘ Longevity Research - LBF7 Spain
This article announces LBF7 Spain, a longevity research event bringing together experts to collaborate on aging solutions.
Key Points:
β’ Gathering top researchers in the field of longevity.
β’ Collaborative event at a longevity resort in the Spanish mountains.
β’ Focus on building practical solutions for extending lifespan.
π Resources:
β’ Long Bio Fellowship β - Application details
β¨ Finance - Unlocking Value in CSS
This article discusses the potential of unlocking value in a little-known company, CSS, through the creation of a US FinTech company.
Key Points:
β’ CSS is a currently undervalued asset of Fannie & Freddie.
β’ A new US FinTech company will hold CSS and offer MBS FinTech services.
β’ This initiative is expected to unlock significant value for the American economy.
π Resources:
β’ Pulte's Twitter β - Shared information about unlocking value in CSS
Image
π‘ Social Media - Algorithm Prioritization Feature Request
This article presents a user request for a feature allowing users to prioritize certain accounts in their social media feeds without enabling notifications.
Key Points:
β’ Request for a "star" function to prioritize specific accounts in the feed.
β’ Prioritization without receiving notifications.
β’ Ideal implementation for following public safety accounts.
π€ Robotics - Data Quality for Robot Learning
This article introduces CUPID, a method for curating robot learning data based on its impact on policy success.
Key Points:
β’ CUPID assesses data quality based on its influence on policy behavior.
β’ Prioritizes data that drives closed-loop policy success.
β’ Moves beyond visual quality to focus on functional impact.
π Resources:
β’ Agia Chris's Twitter β - Shared information about CUPID
Image
β¨ Robotics - Best Paper Award for Long-History Policies
This article announces a Best Paper Award for research on learning long-history policies at the Workshop on Learned Robot Representations.
Key Points:
β’ Research on long-history policies received a Best Paper Award.
β’ Paper available online for review.
β’ Recognition for contributions to the field of robotics.
π Resources:
β’ Marcel TorneΒ΄s Twitter β - Shared the award news and paper link
β’ RoboticsSciSys β - Workshop organizers
β’ Paper β - Research paper on long-history policies
Image
Image
Image
π€ Reinforcement Learning - Diffusion Steering (DSRL)
This article introduces Diffusion Steering Reinforcement Learning (DSRL), an efficient RL method using diffusion models.
Key Points:
β’ Actor chooses noise, denoised by policy to produce actions.
β’ Remarkably efficient RL method for diffusion/flow-based policies.
β’ Improves efficiency of reinforcement learning.
π Resources:
β’ Sergey Levine's Twitter β - Shared information about DSRL
Image
βοΈ Support
If you liked reading this report, please star βοΈ this repository and follow me on Github β, π (previously known as Twitter) β to help others discover these resources and regular updates.
