I'm a Robotics Software Engineer with professional experience working on state-of-the-art humanoid robots. I'm fascinated by the intersection of AI and Robotics, and I'm dedicated to integrating cutting-edge AI advances into practical applications.

I'm currently focused on seamlessly integrating perception with manipulation in an end-to-end model. This repository documents my ongoing efforts in employing the latest transformer innovations to formulate a universal solution for a variety of manipulation tasks. general_manipulation

In the recent MoveIt Hackathon, I secured the 3rd position with my project which bridges image segmentation and object classification. This processed information is then relayed to MoveIt for object grasping planning. Although I couldn't fully explore the grasping phase due to time limitations, this repository provides a comprehensive look into the initial stages of my project. grab_anything

An initiative to develop an agent capable of interacting with the user, and in doing so, creates a model of the individual by reflecting on this interaction. It's equipped with diverse modules like Speech-to-text, text-to-speech, etc., facilitating the agent's communication through hardware and translating it into natural language for tangible device deployment. The interactions are deliberately designed, making the loop deterministic. aware

Following advancements like Auto-GPT, I was inspired to develop a multi-agent LLM system. This framework consists of memory storage, capabilities, and a pivotal task manager. It autonomously plans, focuses on pertinent memory data, and executes the relevant capability as instructed in the docstring. The code is modular and streamlined, though it currently requires adjustments for data storage since its original codebase with Supabase was deprecated. Two potential solutions include local storage or transitioning the data to Weaviate. newrail_bot

• Email: [email protected]
• LinkedIn: Luis Lechuga Ruiz