Skip to content
Drix10 Blog

Acknowledgements - Collaborative Project

, 10 items in AI and Robotics Applications, 5 min read

In this digest (10 items)

This article acknowledges the contributions of various individuals and teams to a collaborative project. It also includes links to relevant profiles and project information.

Key points

• Recognition of Yitong and Guang's contributions.

• Acknowledgement of support from Dale, Paarth, Kuni, Huan Zhang, and Katherine.

• Expression of gratitude to advisor Yunzhu Li for ongoing support and guidance.

Sources

Robotics - Physical AI Presentation

This article summarizes a presentation on Physical AI and intelligent robotics, featuring a discussion with Unitree and Auki Labs representatives. The author represented Persona AI.

Key points

• Presentation on Physical AI and its implications for intelligent robotics.

• Focus on the future of robotics with bipedal locomotion.

• Collaboration with Unitree and Auki Labs.

Sources

Image

Tactile Sensing - 3D-Printable Tactile Sensor

This article introduces eFlesh, a 3D-printable, customizable tactile sensor designed to simplify the integration of touch sensors into robotics.

Key points

• Addresses the difficulty of integrating touch sensors in robotics.

• eFlesh offers 3D-printability and customizability.

• Simplifies the process of adding tactile sensing to robots.

Sources

Image

AI Research - Addressing Research Debt

This article discusses a Stanford talk highlighting the accumulation of "research debt" in the AI community, advocating for a shift towards more robust and less flashy approaches.

Key points

• Emphasis on a Stanford talk about research debt in AI.

• Comparison of research debt to technical debt.

• Call for a move away from flashy demos towards more robust methods.

Sources

Robotics - Overnight Projects

This article describes several projects undertaken in a single night, focusing on gamification, integration testing, and automation in robotics and smart home applications.

Key points

• Gamification of the Altruist onboarding experience.

• Integration tests with a RISC-V smart home server.

• Automation setup for an air purifier based on outdoor sensor data.

Sources

Image

Embodied Agents - GENEA Workshop

This article announces the GENEA Workshop focusing on co-speech generation for embodied agents, with a call for paper submissions.

Key points

• Workshop focuses on co-speech generation for embodied agents.

• Emphasis on the importance of nonverbal communication.

• Paper submission deadline is July 9th.

Sources

Image

Robotics - Happy Hour Event

This article summarizes a successful happy hour event that brought together researchers, engineers, and robotics companies.

Key points

• Networking event featuring DCG, NUS Engineering, and Unitree Robotics.

• Focus on the intersection of deep research, real robots, and human interaction.

• Attendees included builders, deployers, and enthusiasts.

Sources

Image
Image
Image
Image

Robotics - Casper Robot Sidekick

This article introduces Casper, a robot sidekick designed to proactively assist users based on inferred needs and intentions.

Key points

• Casper anticipates user needs and proactively offers assistance.

• Focuses on active sensing rather than passive command reception.

• Maintains user control while providing support.

Sources

Image

Robot Control - Steering Robot Behavior

This article discusses the challenge of controlling powerful robot models trained on large-scale data to align with user intent and deployment needs.

Key points

• Focus on controlling powerful robot models trained on large datasets.

• Importance of aligning robot behavior with user intent and preferences.

• Exploration of methods to steer and control robot behavior effectively.

Sources

Image

Robot Control - Prompting with the Future

This article introduces "Prompting with the Future," an MPC framework combining a pretrained VLM with an interactive digital twin for more effective robot control in open-world scenarios.

Key points

• Focuses on solving tasks requiring semantic and physical reasoning.

• Utilizes a combination of pretrained VLM and interactive digital twin.

• Enables grounded, open-world robot control with limited data.

Sources

Image