Hardware Podcasts

Why humanoid robot demos still fail the generalization test

The Robot Report Podcast · 9 Oct 2026 · 1h 18m

What this episode covers

Janine Sinanan-Singh, Director of Gen AI Research at Appen, explains why humanoid robot demos often fail the generalization test, emphasizing that high-quality egocentric video data and human judgment are critical for training models that can handle out-of-distribution tasks. She discusses the shift from classical robotics to AI-driven approaches, noting that while foundation models like Gemini offer zero-shot capabilities, latency and the need for intent understanding remain significant hurdles. The conversation also covers the role of synthetic data in scaling training, the importance of multi-agent systems for safety, and the economic pressures driving token optimization in AI development.

Most relevant to

early-stage companies

Topics

Mentioned

Companies
Appen, Boston Dynamics, OpenAI, Google DeepMind, Rec, Engine AI, Schneider Electric, PTC, Wibotic, Locust Robotics, Berkshire Grey, Amazon, Hyundai, SoftBank, Siemens
Products
Gemini

Related episodes from other shows

Other shows on the same ground, closest first.

  • The Robo Success Podcast · 20m

    Ep. 17: Today's Humanoid Robots Have Toddler Hands, Not Adult Ones.

    Jay Li, CEO and Founder of Proception, explains that current humanoid robots possess 'toddler-level' dexterity, lacking the tactile sensing required for useful work. He argues that while hardware and range of motion have improved, tactile feedback is now the critical bottleneck for high-success-rate manipulation. Proception addresses this by providing both robotic hands for deployment and specialized gloves for high-quality data collection to train AI models.

    Robotics & autonomySensors & measurement in products
    6 chapters

  • The Robo Success Podcast · 27m

    Ep. 11: What’s Next for Humanoids? Lessons from Tokyo | The Robo Success Podcast

    Hosts Jesica Chavez and Andy Diepen discuss insights from the Humanoid Summit in Tokyo, noting a shift in the humanoid robotics industry from hardware demonstrations to compliance, safety, and real-world deployment. They argue that while the technology is ready, market adoption is hindered by workforce fear of job replacement and a lack of education for end-users. The conversation highlights the critical need for robotics companies to target marketing and customer success at the lowest-level operators to ensure successful integration and ROI.

    Marketing & brandScaling production
    5 chapters

  • EE Times Current · 47m

    Interview: Roboticist Ayanna Howard | Memories May Be Beautiful, And Yet | Math & Neptune

    Georgia Tech professor and roboticist Ayanna Howard discusses the intersection of AI and robotics, highlighting how humans overtrust robots in high-risk scenarios and the challenges of generalizing robotic behavior across different environments. EE Times contributor Gary Hilson analyzes the semiconductor memory market, explaining why emerging memories like MRAM and 3D XPoint struggle to displace established DRAM and Flash due to cost and manufacturing hurdles. The episode also previews the upcoming Boards and Solutions Conference, focusing on embedded computer boards for industrial automation, smart cities, and edge computing.

    Robotics & autonomyChips, FPGAs & semiconductors
    9 chapters

  • The Amp Hour · 1h 18m

    Robo Fry Guy

    Dave Jones and Chris Gammell discuss the feasibility of humanoid robots in low-end service jobs, citing high costs and the 'jack of all trades, master of none' design philosophy, while Jones details Tesla's dense supercomputer power supply failures caused by MEMS oscillator resonance. The hosts also explore practical hardware engineering, including Gammell's use of DALL-E for AI image generation, his 'Lego prototyping' with off-the-shelf sensor boards, and Jones's audit of household phantom power using specialized meters. Finally, they debate the challenges of sourcing standard plastic enclosures and the logistics of remote hardware training via browser-based containers.

    AI & machine learningRobotics & autonomy
    8 chapters

  • The Next Byte Podcast · 1h 02m

    192. Karsten Roscher & The Need For Safe AI (Interview)

    Karsten Roscher, Head of the Certifiable AI department at the Fraunhofer Institute for Cognitive Systems (IKS), discusses the critical need for safety in AI development, particularly in autonomous driving. He explains the challenges of out-of-distribution detection, where models fail on unseen data like e-scooters or horse carriages, and introduces Robuscope, a free tool for evaluating model robustness without sharing proprietary data. Roscher also critiques the industry's over-reliance on benchmarks and highlights a new SAE International course designed to bridge the gap between machine learning engineers and safety experts.

    AI & machine learningDebugging & failure analysis
    6 chapters

The summary, chapters and topics are generated from the episode’s audio by this directory, not written by The Robot Report Podcast. The episode itself belongs to its publisher.