🤖 F.02 Walks Naturally Like a Human Now

Subscribe | AI Toolkit | Meet The Team

Welcome back AI enthusiasts!

In today’s Daily Report:

• 🦿F.02 Walks Naturally Like a Human Now

• 🦫Alibaba’s New AI Can See, Hear, Talk, and Write!

• 🛠Trending Tools

• 🥪Brief Bites

• 💰Funding Frontlines

• 💼Who’s Hiring?

Read Time: 3 minutes

🗞RECENT NEWS

FIGURE AI

🦿F.02 Walks Naturally Like a Human Now

Figure AI developed “Learned Natural Walking (LNW),” which enables Figure 02 (F.02) to walk naturally like a human.

Key Details:

• F.02 is a general-purpose humanoid robot optimized for manufacturing, warehousing, and household environments.

• “LNW” relies on Reinforcement Learning (RL), which teaches F.02 to learn the optimal behavior in an Environment to receive Positive Rewards.

• RL is comprised of four components:

  1. Learner: The humanoid robots.

  2. Environment: The simulated real-world settings in which the humanoid robots are trained in.

  3. Policy: The instructions the humanoid robots follow to take action.

  4. Feedback: The Positive Rewards or Negative Penalties the humanoid robots receive after taking action.

• Positive Rewards are given when the humanoid robots successfully maintain balance and recover from trips, slips, and shoves.

• Negative Penalties are given when the humanoid robots fall, lose balance, or exhibit unnatural movements.

Why It’s Important:

• “LNW” enables F.02 to train on years of simulated data in only a few hours with an AI-powered Sim-to-Real Transfer (S-t-RT) technique.

• When humanoid robots train in simulated environments, they often struggle to perform in real-world settings because of differences in physics, friction, and velocity. S-t-RT accounts for these differences.

🩺 PULSE CHECK

AI RESEARCH

🦫Alibaba’s New AI Can See, Hear, Talk, and Write!

Alibaba Cloud just released “Qwen2.5-Omni-7B,” a new multimodal AI model that simultaneously processes text, images, audio, and video on personal devices like smartphones or laptops.

Key Details:

• “Qwen2.5-Omni-7B” leverages two components:

  1. Thinker-Talker Architecture (T-TA): Thinker processes and understands text, images, audio, and video inputs to generate summaries of what they mean. Talker turns these summaries into natural-sounding speech outputs.

  2. Time-aligned Multimodal RoPE (TMRoPE): Synchronizes video inputs and audio inputs. For example, making sure dialogue from a movie scene matches the actor’s lip movements.

• T-TA and TMRoPE allow for real-time translation of text inputs and real-time audio descriptions of image inputs or video inputs with minimal latency.

• “Qwen2.5-Omni-7B” achieved 93.4% accuracy on the Seed-TTS-Eval benchmark, which assesses the ability of multimodal AI models to perform Text-to-Speech (TTS).

Why It’s Important:

• Running complex multimodal AI models on personal devices reduces reliance on cloud computing, which improves a user’s privacy and lowers latency.

• By seamlessly integrating multiple modalities, “Qwen2.5-Omni-7B” enables AI to understand, process, and respond to inputs in a more human-like way.

🛠TRENDING TOOLS

✂️ClipZap clips, edits, and translates videos automatically.

🏦Smart Clerk turns bank statements into financial reports.

🎙️Audioread turns articles, PDFs, and emails into podcasts.

🗣️Vocode is an open-source library for building Voice Agents.

🗺️Gitmind generates free AI-powered Mind Maps and Flows Charts.

🔮Browse our always Up-To-Date AI Tools Database.

🥪BRIEF BITES

Amazon launched “Interests,” a new AI-powered feature that finds products that match your passions and hobbies.

A California Federal Judge rejected music publisher UMG’s request to block Anthropic from using song lyrics to train conversational chatbots.

The U.S. Government just added over 50 Chinese Tech Companies to an Exports Blacklist that prevents them from acquiring cutting-edge GPUs.

OpenAI CEO Sam Altman announced that OpenAI will adopt Anthropic’s open-source Model Context Protocol (MCP), which enables ChatGPT to connect to external data sources.

💰FUNDING FRONTLINES

• Nexthop AI closes a $110M Funding Round to build AI Infrastructure for Cloud Companies.

• Lumber lands a $15.5M Series A for an AI-Powered Construction Workforce Management Platform.

• Arlo raises a $4M Seed Round to deploy an AI-Based Health Insurance Underwriter for small businesses.

💼WHO’S HIRING?

• BigID (Remote): Software Integration Engineer Intern, Summer 2025

• Salesforce (San Francisco, CA): Data Scientist, MTS, Entry-Level

• Airtable (New York, NY): Data Scientist, Finance Analytics, Mid-Level

• Plus (Santa Clara, CA): Senior Embedded Software Engineer, Senior-Level

📒FINAL NOTE

FEEDBACK

❤️TAIP Review of The Day

REFER & EARN

🎉Your Friends Learn, You Earn!

You currently have 0 referrals, only 1 away from receiving ⚙️Ultimate Prompt Engineering Guide.

Copy and paste this link to friends: https://theaipulse.beehiiv.com/subscribe?ref=PLACEHOLDER