Xpeng's flying car taking off

Good morning, robotics enthusiasts. Humanoids and flying cars are set to take off soon, as China’s Xpeng gears up to mass-produce both by 2026.

Flying cars like Xpeng’s are crossing over from sci-fi to reality, but questions remain: Will they transform public transport or just become air-clogging playthings for the uber-rich? What do you think?

In today’s robotics rundown:

• Xpeng to ‘mass-produce’ flying cars

• Humanoids that can smell like humans

• The mind-controlled robot arm

• MIT nudging robots to fix mistakes

• Quick hits on other robotics news

LATEST DEVELOPMENTS

XPENG

🛸 Xpeng to ‘mass-produce’ flying cars

The Rundown: Chinese EV maker and Tesla rival Xpeng just announced plans to mass-produce humanoids and its “modular flying car” by 2026—news that boosted its shares to their highest level since August 2022.

The details:

• Xpeng will scale commercial production of its humanoid, Iron, next year, equipping it with Level 3 autonomy—requiring oversight in select scenarios.

• The flying car in question is Xpeng’s Land Aircraft Carrier, an electric minivan with an eVTOL vehicle, or passenger drone, tucked in the back.

• Priced at 2 million yuan ($275,000), the hybrid vehicle accommodates four passengers in the van portion and two in the detachable drone.

• Xpeng said that it has so far secured around 3,000 orders, with potential customers focusing largely on tourism and short-haul rescue missions.

Why it matters: Xpeng has already deployed Iron in its EV factory and begun construction on a plant designed to produce 10,000 flying cars annually. While U.S. rivals like Beta Technology advance their eVTOLs, Xpeng stands to benefit from China’s ambition for the “low-altitude economy,” which is set to reach $479B by 2035.

AINOS

🌸 Humanoids that can smell like humans

The Rundown: San Diego-based biotech and research company Ainos announced a novel partnership with Japanese robotics company ugo to develop the “world’s first robots” with a sense of smell.

The details:

• The partnership will integrate Ainos' AI Nose tech, an electronic olfactory system that mimics human smell, with ugo's autonomous robotic platform.

• A decade in the making, Ainos’ AI Nose has been found to detect and analyze 22 volatile organic compounds (VOCs) with nearly 80% accuracy.

• The move is expected to revolutionize a variety of sectors, from industrial safety to healthcare, with deployments expected as early as 2026.

• Ainos says its AI Nose can analyze exhaled breath to identify abnormalities and potentially diagnose conditions, including cancer and respiratory diseases.

Why it matters: While robot perception has traditionally been limited to vision, sound, and touch, the digitization of smell aims to change that—unlocking numerous potential applications. AI Nose-equipped robots could soon detect gas leaks in factories, screen public spaces for explosives, and even sniff out diseases in medical settings.

UCSF

🦾 The mind-controlled robot arm

The Rundown: Researchers at UC San Francisco (UCSF) achieved a groundbreaking milestone in brain-computer interface (BCI) technology, enabling a paralyzed man to control a robotic arm using only his thoughts.

The details:

• The system uses tiny sensors implanted on the surface of the participant's brain to detect neural signals when they imagine specific movements.

• An AI model then decodes these signals and translates them into precise commands for the robotic arm to execute.

• The study participant, who was paralyzed due to a stroke, trained the system over two weeks by imagining finger and hand movements.

• After transitioning to the robotic arm, he successfully performed tasks such as opening a cabinet, retrieving a cup, and placing it under a water dispenser.

Why it matters: Traditional BCIs often degrade in performance within days due to neural drift, requiring frequent recalibration. However, UCSF’s AI-powered system adapts to subtle shifts in brain activity, maintaining functionality for seven months. This brings BCIs closer to real-world adoption by making them more practical and reliable.

MIT

🤖 MIT nudging robots to fix mistakes

The Rundown: MIT and NVIDIA researchers developed a new AI framework that allows users to correct a robot’s behavior with simple interactions, such as nudging a robot’s arm, rather than retraining its machine-learning model.

The details:

• The new framework, detailed on arXiv, doesn’t require users to collect new data or fine-tune a neural network to fix new (small or big) mistakes.

• Instead, users correct the robot’s behavior through simple, intuitive interactions like pointing, tracing trajectories, or physically guiding its arm.

• In tests, the method achieved a 21% higher success rate compared to alternatives that didn't leverage human interventions.

• It aims to make robots user-friendly right out of the box and ultimately transform human-robot interaction across areas like healthcare and elder care.

Why it matters: MIT’s research joins models like TRACE and ACORD in enhancing real-time interaction between humans and robots—key to improving their effectiveness, safety, and usability. The approach helps bridge the gap between pre-trained robots and real-world deployment.

QUICK HITS

📰 Everything else in robotics today

Bank of America forecasts that the world will host 3 billion humanoids by 2060, with 65% being in homes, 32% in services, and 3% powering industrial applications.

Japan’s service robot market is projected to triple in five years to $2.7B, driven by labor shortages, with major restaurant chain Skylark already using 3,000 robots.

U.S. researchers found that people trust humanoids significantly more than nonhumanoid ones to care for objects, personal information, children, and pets.

German researchers developed a biorobotic arm with two strands of artificial muscles attached to the forearm, designed to suppress tremors.

Chinese EV maker BYD unveiled a vehicle-mounted drone launching system, developed in collaboration with DJI, that is available for all of the company’s vehicles.

Lifestyle robotics company Aescape, makers of AI-powered massage robots, raised $83M in funding, taking its total capital raised to $128M.

Defense aircraft provider Shield AI secured $250M in funding to expand its Hivemind platform for autonomous drones and robots, raising its valuation to $5.3B.

German researchers are currently developing new robot grippers that require 90% less electricity than conventional systems.

COMMUNITY

🎥 Join our next live workshop

Join our next special workshop this Tuesday at 3 PM EST to learn about how AI can make you a full-stack developer using Bolt AI prompts —whether you’re just starting out or looking to level up your coding skills, with Tomek Sułkowski, The Bolt.new’s founding engineer.

RSVP here. Not a member? Join The Rundown University on a 14-day free trial.

See you soon,

Rowan, Jennifer, and Joey—The Rundown’s editorial team