Can a toy robot feel real? Cozmo says yes

ADVANCEMENTS IN ROBOTICS

1. 🤖Hugging Face’s Reachy2 - The open-source humanoid that understands you:

Imagine training a robot just by talking to it.

Hugging Face and Pollen Robotics have launched Reachy2, an AI-native, open-source humanoid robot that marks a serious leap in human-machine interaction.

What’s revolutionary?
👉 It’s powered by Large Language Models (LLMs) like those behind ChatGPT.
👉 You can teach it new tasks in plain English, no complex programming required.
👉 It’s open source from its physical blueprint to its software stack.

Why it matters?
For the first time, robotics is being built like open software community-driven, transparent, and modular. Startups, researchers, and even hobbyists can now tinker, test, and improve humanoid capabilities without enterprise budgets.

Think of Reachy2 as the Raspberry Pi moment for humanoids.

This democratization of robotics will lower the entry barrier, encourage creative experimentation, and accelerate real-world use cases from elder care to warehouse work.

AI was once just about digital intelligence. Reachy2 adds arms, hands, and eyes to it.

2. Sensory: Robots can now feel the difference between a pat and a punch, thanks to KAIST’s neuromorphic “artificial nerves.” Like built-in instincts, these chips let machines ignore gentle touches and react instantly to threats boosting energy efficiency and responsiveness in tactile robotics.

3. Believable: A playful little robot named Cozmo has revealed how simple game-like interactions spark our brains into seeing machines as sentient. The study from UEA shows that social cues can trigger a sense of agency asking us: if a toy “plays” with us, could it truly feel alive?

4. Simulated: MIT’s CSAIL just unleashed PhysicsGen, a simulation pipeline that turns a few VR demos into thousands of robot training routines. The result? Robotic arms learn tasks with 30% higher success, pointing toward foundation-model-style intelligence for bots.

5. Automated: Hyundai has deployed German-built humanoid welders in shipyards, with prototypes due by end‑2026 and rollouts expected in 2027. These bots are precision-engineered to boost safety and fill skilled-labor gaps in heavy industry.

ROBOTICS ON SOCIAL MEDIA THIS WEEK

1. 🧱Brick Bot: Watch LEGO bricks transform into a Spider Hero robot in this fun build.

2. 🤖Robo Reel: Dive into this week’s coolest robot videos from smart structures to underwater tech.

3. 🤖Art Bot: World’s first humanoid robot artist paints King Charles III in stunning ultra-realism.

4. 🤖Groove Bots: Tiny robots EMO & Eilik show off epic dance moves in perfect sync.

5. 🗣️Talk Bot: Meet Reachy Mini, your desktop robot that listens, talks, and connects with you.

6. ⚙️Evolved: Robots are now running, adapting, and working like never before. The future is no longer sci-fi.

SUNDAY SCHOOL

AI Video Generator: Text to Video in Seconds!

SPECIAL

Robots that learn like humans, just swap tools and go

A team at TU Delft just cracked a major robotics bottleneck: teaching robots to reuse skills, instantly.

In their latest research, robots were trained to perform tasks like opening doors or pouring drinks. The twist?
You can swap tools, and they automatically adapt to the new object with zero retraining.

How?
🔧 Using 3D learning and motor control transfer.
🤖 A robot trained to pour water with a cup could now do it with a teapot without starting from scratch.

Why it’s a game changer?
Today, robotic training is painfully slow and specific. But this method enables robots to generalize their skills, similar to how humans transfer muscle memory between tools.

For manufacturing, logistics, or home robotics, this means:

▪️ Faster deployment

▪️ Lower cost of development

▪️ Smarter, more flexible machines

Combined with LLMs and open-source robots like Reachy2, we’re inching closer to general-purpose robots that don’t just follow instructions, they adapt.

This is more than automation. It’s evolution.

PROMPT OF THE WEEK

Write a cheat sheet for financial statement notes disclosures?

Act as a professional financial analyst and reporting expert with deep knowledge of accounting standards (e.g., IFRS or Indian GAAP).

Your task is to create a concise, easy-to-reference cheat sheet listing the most common notes to financial statements and what each typically includes. This cheat sheet should be useful for finance teams, auditors, or students reviewing financial reports.

✅ FORMAT TO FOLLOW:
List 10–15 of the most common disclosures with:
- Title of the Note (e.g., Revenue Recognition, Property Plant & Equipment)
- 1-line purpose — Why it’s included
- Bullet points of typical content — Key components that must be disclosed

🧾 CONTEXT:
- Audience: Junior accountants, CA students, finance team members, early-stage founders
- Tone: Clear, professional, and practical
- Framework: Use a general structure applicable to Indian companies, but keep it adaptable to global accounting practices

📌 EXAMPLES TO INCLUDE:
- Accounting Policies
- Revenue Recognition
- Property, Plant & Equipment (PPE)
- Contingent Liabilities
- Related Party Transactions
- Financial Instruments
- Share Capital
- Income Taxes
- Provisions & Contingencies
- Earnings per Share (EPS)
- Leases (Ind AS 116 or IFRS 16)

The final output should read like a handy internal reference — simple enough for quick understanding, but accurate enough for use in real financial review work.

If you find the contents of this email useful, subscribe now & share with your friends.

Your Opinion Matters!

Got more feedback or just want to get in touch? Reply to this email and we’ll get back to you.

Thanks for reading.

Until tomorrow!

Shinky & the Hanoomaan AI team