Tesla Rival Figure AI's Humanoid Bots About To Go BALLISTIC

Figure AI, a company developing humanoid bots, is poised to make a major announcement showcasing their capabilities, with potential applications in various industries, and may be on the verge of revolutionizing labor management and logistics with their rapidly advancing technology

Questions to inspire discussion

Training and Capabilities

🤖 Q: How quickly can Figure's humanoid robots be trained for complex tasks?
A: Figure's robots can be trained in tens of hours using the Helix parameter model to perform tasks like dishwasher loading, towel folding, and package sorting.

🧠 Q: What is a key challenge in training these robots for longer-term tasks?
A: Long horizon planning is more challenging than short-term tasks, requiring the robots to adapt to different objects and materials over extended periods.

🔄 Q: How do the robots learn to generalize and perform new tasks?
A: The robots can recognize similar objects and patterns to perform new tasks, even if not explicitly trained on them, such as putting ketchup in the fridge based on shape and size recognition.

Deployment and Applications

🏨 Q: Where can these robots be initially trained before home deployment?
A: Figure AI's robots can be trained in homelike settings such as hotels, hospitals, and restaurants to learn from diverse tasks and environments before individual home use.

🧺 Q: What are some specific environments where these robots can be trained?
A: The robots can be trained in settings like laundromats for laundry folding and gardens for watering plants, providing diverse training data for adaptation to new situations.

👥 Q: How can businesses utilize these robots for flexible labor solutions?
A: Companies can use services like Salesforce's robot as a service to request specific tasks and robots for deployment at specific times, managing robot fleets for on-demand labor.

Integration and Features

🗣️ Q: What additional capabilities can Figure's robots gain through integration with Salesforce?
A: Integration with Salesforce's AI workforce (Agent Force) enables voice communication and selfie-taking capabilities, allowing more natural human interaction.

⏱️ Q: What is a key technical issue that needs improvement in these robots?
A: The latency or response time of Figure's robots, currently at 5-6 seconds, needs to be reduced by improving edge compute and connectivity.

Industry Impact and Challenges

🏭 Q: In which industries can Figure's humanoid robots be deployed?
A: These robots can be deployed in industries including manufacturing, customer service, and elder care, leveraging their human-like interaction capabilities.

📈 Q: What is a major challenge for the widespread adoption of these robots?
A: Scalability is a key challenge, with the robots' complexity and training set size being crucial factors in determining their potential applications and scalability.

🤲 Q: What are the primary challenges in improving the robots' manipulation skills?
A: Dexterity, hands, and touch are key challenges in enabling the robots to manipulate various objects and materials effectively.

🏆 Q: How does Figure's progress compare to other companies in the field?
A: Figure's rapid advancements put their humanoid robots head-to-head with Tesla's Optimus in the humanoid robotics race.

Key Insights

Rapid Training and Adaptability

1. 🤖 Figure Robotics' humanoid robots can be trained in tens of hours using the Helix neural network system to perform tasks like dishwasher loading, towel folding, and package sorting.
2. 🧠 The robots can learn and generalize tasks they weren't explicitly trained for, such as putting ketchup in the fridge, by recognizing similar shapes and sizes.
3. 🏨 Training in homelike settings like hotels, hospitals, and restaurants with uniform materials provides the diversity needed for adaptation.

Dexterity and Manipulation

1. 🖐️ The dexterity of Figure's robots is a significant achievement, with fingers able to manipulate dishes and towels with ease.
2. 🧺 Robots can adapt to different objects like saucers, cups, and glasses, as well as soft materials like towels, but may struggle with varying fabrics or patterns.

Advanced Applications

1. 🤝 Figure's robots can be used for customer service and hospitality, including taking selfies and providing information at events like Dreamforce.
2. 🔗 Integration with Salesforce's Agent Force AI platform enables more advanced and personalized customer services.

Challenges and Future Developments

1. 📊 The long horizon problem of loading an entire dishwasher, requiring planning and strategy, is a real challenge in training humanoid robots.
2. 🏭 Humanoid robots may not be ready for individual use until they can be produced in millions, according to Scott Walter.

Autonomy and Scalability

1. 🔄 Figure's humanoid robots are fully autonomous but still require training and guidance from human operators to learn new tasks.
2. 📈  The scalability of these robots is key, as they become more advanced and capable of a wider range of tasks and applications.

Competition in Humanoid Robotics

1. 🏁 Figure AI's robots are now head-to-head with Tesla's Optimus in the humanoid robotics race, signaling potential readiness for certain commercial situations.

#CobotsXMentions: @HabitatsDigital @RoydenDSouza @GoingBallistic @AnatomyUMea

Over The Horizon Url: https://www.youtube.com/watch?v=QmaXyUmi0d0

Clips

• 00:00 🤖 Figure AI, a Tesla rival, may make a major announcement around October 6th, showcasing their humanoid bot's capabilities, including interaction and selfies, potentially tied to a movie or Salesforce event.
  • The discussion centers around Figure Robotics, led by CEO Brett Atcock, and recent posts from Atcock and Salesforce CEO Mark Beni off, analyzed by robotics experts Scott Walter and Gustav Anderson.
  • The conversation revolves around the movie Blade Runner, specifically referencing its release date of October 6, 2021.
  • The speaker speculates that a potential announcement from Figure AI, a Tesla rival, may be connected to the US release date of a movie or a Salesforce event around October 6th or 14th.
  • Figure AI's humanoid bot will be showcased at Dream Force in San Francisco, allowing attendees to interact with it, ask for selfies, and receive information.
  • A conversation between a human and a Figure AI robot demonstrates the robot's capabilities, including taking a selfie and providing information about a Salesforce event.
  • Mark Benov released a video showing Figure AI's humanoid bot with speech powered by Salesforce's Agent Force, an integration that sounds natural but may not be fully native or optimized yet.
• 06:13 🤖 Figure AI's humanoid bots, backed by Salesforce, are poised for large-scale deployment in various industries, potentially transforming labor management and logistics.
  • Latency in communicating with humanoid bots, such as those developed by Figure AI, may be reduced by increasing edge compute and improving connectivity.
  • The developers of Figure AI's humanoid bots were likely caught off guard while working on a development build, and the premature video release was probably not intended for public consumption.
  • Salesforce is likely preparing to invest in and deploy humanoid robots, possibly through its venture capital firm, as hinted by a recent white paper on robotics and the concept of a "physical" Agent Force.
  • Figure AI's humanoid bots, backed by Salesforce, may integrate with CRM systems to help companies manage labor requirements, leveraging Salesforce's fleet software and knowledge.
  • Figure AI's humanoid bots may not be ideal for every application, allowing smaller companies to offer specialized solutions, such as logistics robots with wheels, that better suit specific needs.
  • Embodied AI in humanoid bots is being deployed in various industries, particularly in China, for simple tasks, and this trend is expected to increase, with more deployments happening in factories and warehouses, including Amazon, over the next year.
• 13:15 🤖 Figure AI's humanoid bots are being developed for various use cases and are quickly learning tasks, such as autonomously loading a dishwasher, with potential applications in elder care, security, and more.
  • Figure AI's humanoid bots have various use cases, including elder care, hospitals, and hospitality, and can be used for security patrols, such as monitoring facilities and investigating suspicious activity.
  • Figure AI's humanoid bot, using Helix, was quickly taught to autonomously load a dishwasher, building on previous tasks like unloading groceries, laundry, and package sorting.
  • Figure AI's humanoid bot, Helix, has been trained for only tens of hours with a large parameter model and a growing training set, with more development still needed.
  • Figure AI's humanoid bots have learned to adapt and manipulate objects, such as picking up dishes, by applying techniques like applying pressure to make it easier to grasp.
  • Figure AI's humanoid bots are learning tasks quickly, starting with simple, replaceable items, and will eventually handle complex tasks, similar to how full self-driving technology will need to address numerous edge cases.
• 17:20 🤖 Figure AI's humanoid bots show impressive dexterity and are being developed to handle complex tasks, with challenges remaining in handling soft objects and long-horizon problems.
  • Figure AI's founder reflects on the rapid progress made by the company since its launch three years ago.
  • Figure AI's humanoid bots demonstrate impressive dexterity, such as manipulating objects with two hands, showcasing neural networks' potential for complex tasks.
  • The biggest challenge for Figure AI's humanoid bots is handling soft, flexible objects like towels, which is harder than dealing with rigid objects, and also determining the best approach for complex tasks.
  • Loading a dishwasher is a long-horizon problem for AI systems, requiring longer-term planning and strategy to execute, unlike simple tasks like grabbing objects.
• 21:39 🤖 Figure AI's humanoid bots may struggle with tasks due to color/pattern recognition issues, but aim to improve with reliable actuator feedback and potential affordability for individuals through mass production.
  • The AI's humanoid bots may struggle with tasks if they are taught to recognize specific colors or patterns, as changes in these variables can cause them to fail.
  • Training humanoid bots to perform tasks, such as handling objects, can be frustrating and may not require perfect tactile information, but rather reliable actuator feedback.
  • Figure AI's humanoid bot, likely to be sold to companies first under a Robot-as-a-Service (RAS) model, may eventually be affordable for individuals once mass production is achieved.
• 24:26 🤖 Figure AI's humanoid bots are being rapidly developed for widespread use in various settings, with potential to revolutionize industries like hospitality and healthcare.
  • Figure AI's humanoid bots will start training in homelike settings such as hospitals, hotels, and restaurants where materials and sizes are standardized.
  • Figure AI's humanoid bots and other startups' simpler robots are being rapidly developed and deployed for various tasks, such as laundry and hotel room cleaning, with increasing capabilities and potential for widespread use.
  • Figure AI's humanoid bots can be trained quickly for specific tasks, such as navigating hotel rooms, making them potentially useful for various applications.
  • Figure AI's humanoid bots can generalize and adapt to new objects similar to those they've been trained on, as demonstrated by their ability to put away objects in a kitchen without prior explicit training.
  • Figure AI's humanoid bot can identify and interact with new objects, such as putting a ketchup bottle in the fridge, by generalizing from similar items it has seen before.
  • Salesforce may become a labor-on-demand service, managing a fleet of humanoid bots from various manufacturers to provide temporary robotic labor for commercial situations.
• 30:16 🤖 Figure AI's humanoid bot reveal, potentially with personalized AI, is speculated to happen before October 14th, possibly rivaling Tesla.
  • The speaker analyzes a cryptic image posted by Brett, speculating it may be related to upcoming announcements from Figure AI, but its meaning remains unclear.
  • The speaker believes Figure AI's F3 humanoid bot reveal will likely occur before October 14th, specifically before Salesforce's Dreamforce event.
  • The speaker speculates that Figure AI's humanoid bots may gain a personalized AI or androgynous personality by the end of the third quarter, possibly around August or October.
  • Figure AI's humanoid bot, Figure One, had a voice demo speculated to be from a celebrity, ultimately revealed to be Jared Leto, indicating some Hollywood actors are willing to lend their voices to AI projects.
  • The conversation ends with speculation about Figure AI's humanoid bots, with no concrete information provided.
• 36:02 🤔 The conversation with Scott and Gustav ends with a promise to speculate further next time.

-------------------------------------

Duration: 0:36:25

Publication Date: 2025-09-16T19:04:02Z

-------------------------------------