Introduction

> Hardware innovation is crucial in robotics. It's not enough to focus on software alone; the physical components need continuous advancements. "People who think you don't need to innovate hardware anymore are wrong."

> The natural movement of our robots, whether walking, running, or flipping, is a key differentiator. We've pushed our creations to carry incredible loads, like "having one carrying the other one," to ensure robustness and reliability.

Early robots

> My fascination with robotics really took off during my time in graduate school at MIT, when I saw a Vyke robot arm disassembled into a thousand pieces. That moment sparked my imagination profoundly, and I realized how robotics could transcend my studies in neurophysiology by offering a more conceptual approach to understanding intelligence and control systems.

> Throughout my career, the balance between function and creativity has been pivotal. Initially, I believed function was paramount—mobility, dexterity, perception, and intelligence defined robotics. However, I eventually realized that aesthetics play a crucial role too. It's not just about what robots can do but also how they capture our imagination through their design and movements, adding a layer of lifelike intrigue.

Legged robots

> When I first saw a slow, tripod-stable six-legged robot, I knew it was wrong. People and animals move by bouncing and flying, not cautiously. That inspired me to create hopping robots that utilize springiness and energy circulation like humans do.

> In robot manipulation, we need to break free from static grasping and aim for more aggressive and human-like interactions. Instead of carefully positioning and grasping objects based on geometry, we should juggle, hug multiple objects, and adapt to dynamic situations for more realistic manipulation abilities.

> Finding the right people to work with was crucial. I believe the smartest thing I ever did was to find excellent engineers and collaborators who could bring my dream of creating innovative robots to life while I focused on dreaming and concept creation.

> Initially, I wasn't enthusiastic about humanoid robots due to the focus on form over functionality. But working on humanoids like Atlas showed me the impact of connecting with others. Even though they may not look exactly like humans, the social interactions they enable are compelling and special.

> Starting with physics-based simulations, we worked on projects like the Aibo runner and QRIO, leading us back into the realm of robotics before creating Boston Dynamics. Those projects helped us rediscover our passion for robots and laid the foundation for the innovative work we continue to do today.

Boston Dynamics

> Reflecting on those early days, it was exhilarating to create a surgical simulator that combined robotics with force feedback, allowing surgeons to virtually feel and manipulate tissues. “Surgeons were practically lined up,” driven by competition, but we quickly realized our market wasn’t as expected, leading us to make a tough but essential pivot in our journey.

> There’s something invigorating about recognizing when something isn’t working and having the courage to let it go; it became a defining moment for us. “It just always felt right once we did it,” and that decision to focus on building robots not only refined our vision but also aligned us more closely with our core capabilities.

BigDog

> The transition from a simulation to a robotics company was fueled by early projects like the Aibo runner and collaborations with Sony, despite technical and communicative challenges in a pre-internet era. Working closely with Sony marked a pivotal shift in our focus towards tangible robotics.

> The development of BigDog was groundbreaking for Boston Dynamics, integrating all key components on one platform, allowing us to "take it out in the woods." This project was vital not only in terms of engineering but also in learning how to manage real-world dynamics, leading us to crucial insights about robotics in uncontrolled environments.

> The progression from BigDog to LS3 and eventually to Spot represented a significant evolution in robotics design, especially in moving from hydraulic to all-electric powered systems. This shift was largely inspired by input from visionary leaders, such as Larry Page, who challenged us to create smaller, more human-friendly robots, thus shaping the future of our robotic advancements.

Hydraulic actuation

> - Hydraulics is a technology that I've always had a fondness for. It's often seen as old-fashioned and messy, but its strength-to-weight ratio is impressive. We innovated in areas where there hadn't been much change, like designing smaller, more efficient valves for our robots. Our hydraulic power supply in Atlas, for example, is compact, weighing just five kilograms and producing five kilowatts of power.

> - We made significant advancements in hydraulic technology, introducing new circuits and integrated components, resulting in a powerful yet compact system. Despite its perceived limitations, there is still untapped potential for hydraulics to evolve and offer more innovative solutions in robotics.

Natural movement

> Creating natural, fluid movements in robots like those at Boston Dynamics involves advanced hardware and dynamic algorithms. Instead of relying solely on reactive control, we predict future motions and make real-time adjustments, enabling actions that appear natural and human-like, such as somersaults or running.

> Simulating human movements in robots often reveals the complexities of our own motion. Despite not fully understanding the biomechanics of actions like walking or flipping, robotic replication helps bridge the gap. This journey is as much about advancing robotics as it is about gaining insights into human mechanics.

> Achieving tasks like fast, agile movement in robots requires innovative engineering. For instance, Wildcat, our quadruped robot, could sprint up to 19 miles per hour, albeit loudly, showcasing our philosophy that "you have to run before you can walk." This highlights the intricate balance of speed and efficiency necessary to push the boundaries of what's possible in robotics.

Leg Lab

> Balancing Simplicity and Functionality: When designing robots, it’s crucial to strike a balance between practical constraints and desired outcomes. For instance, our early pogo stick design was an intentional simplification aimed at capturing essential aspects of human locomotion without replicating it exactly. Starting simpler can pave the way for more complex iterations in the future.

> Inspiration from Nature: Observing and analyzing animal locomotion has profoundly influenced our robotic designs. From dissecting ostrich legs to studying the graceful, high-speed movements of cheetahs, these natural models provide invaluable insights. Each animal's unique movement, such as the pronghorn's pronk, offers lessons in efficiency and adaptability that we can translate into more effective robotic systems.

> The Concept of Passive Dynamics: A great robotic design often incorporates passive dynamics, where mechanical systems can perform movements with minimal to no computational control. This concept, championed by Tad McGeer, underscores that the body itself plays a crucial role in motion, not just the brain or computational controls. Efficient and graceful robot movements emerge when the mechanical and computational elements are harmonized, leveraging the natural tendencies of the machine's structure.

AI Institute

> Athletic intelligence and cognitive intelligence are the two crucial parts of intelligence that we aim to integrate at the Boston Dynamics AI Institute. Athletic intelligence involves robotic capabilities like mechanical design, real-time control, and energetics, whereas cognitive intelligence is about planning and understanding, which most robots currently lack. By combining these, we're working on ambitious goals like enabling robots to understand and replicate human tasks through observation, essentially an on-the-job training paradigm for robots.

> The concept of athletic intelligence is a fascinating domain where we overlook how skilled humans are with basic tasks like walking, using hands, and manipulating objects. This innate human capability is not yet fully replicated in robotics, highlighting the complexity and potential for innovation. Achieving a deeper understanding and enhancement of this athletic intelligence in robots remains a challenging and exciting frontier for us.

Athletic intelligence

> The biggest challenge in robotic athletic intelligence is overcoming cognitive limitations, as "the lack of the cognitive side is probably the biggest barrier" to commercial viability and productivity, despite advancements in physical capabilities.

> Progress is most effectively achieved through "stepping stones to moonshots," where we focus on tangible short-term results that inform and motivate long-term goals; successful projects provide valuable feedback, akin to the approach at Boston Dynamics.

> Our aim is to develop robots that can operate successfully in uncertain environments without explicit instructions, reflecting true intelligence by being able to adapt and navigate tasks, much like how we learn through experience rather than detailed explanations.

Building a team

> Building a great team hinges on key qualities: technical fearlessness, diligence, intrepidness, and fun. "Technical fearlessness means being willing to take on a problem that you don't know how to solve," and learning from iterative steps towards a groundbreaking solution is crucial. Each trait empowers the team to tackle unprecedented challenges with creativity and persistence.

> The frontier of robotics lies in creating systems that mimic human cognitive and physical problem-solving abilities. For example, envisioning robots that not only gather data, like Boston Dynamics' Spot robots, but also interpret, diagnose, and fix issues autonomously. This integration of AI with physical capabilities heralds a future where robots could become adept at intricate tasks such as repair and maintenance, shaping a transformative era in robotics.

Videos

> - Diligence is crucial in robotics because it encompasses not just the technical challenges, but also the real-world unpredictability that machines will face. "Having an approach that leaves you unsatisfied until you've embraced the bigger problem" is what drives robustness in robotic systems.

> - Demonstrating both successes and failures in videos is essential to understanding the capabilities and limitations of robots. "Showing both failure and success makes you appreciate the success," as it captures the essence of the engineering process.

> - Embracing intrepidness is vital. Robotics is inherently challenging, and persistence through setbacks is part of the journey; "you have to stick to it and keep trying" to make meaningful advancements in this field.

Engineering

> Engineering is about "having technical fun" and finding satisfaction in creating things you love. Being part of a team with diverse skills adds to the fulfillment. Plus, getting paid for it all makes engineering truly rewarding.

> The essence of engineering is merging science with artistry, going beyond just describing to actually creating. Being at the cutting edge allows for artistic expression and the joy of creating something entirely new. The magic of bringing metal or machines to life is a special kind of fun and creativity.

Dancing robots

> The beauty of robotics lies in "having the robots move in a way that's evocative of life." It's exhilarating to explore elegant movements and the potential for robots to perform high-energy acts, even if we're not competing directly with humans just yet.

> In dance, the vision of a professional ballerina teaching a robot is captivating—imagine the "process of a clueless robot trying to figure out a small little piece of a dance." This interaction could unveil remarkable possibilities in human-robot communication through movement, transcending mere mechanical performance and capturing the essence of expression.

Hiring

> Creating a vibrant engineering culture is crucial; "if you have an environment where interesting engineering is going on, then engineers wanna work there." This realization took years at Boston Dynamics, where we initially attracted passionate individuals who loved building, even if they didn't have professional degrees.

> Robotics has the power to evoke a wide array of emotions and inspire future generations; "the full spectrum of human emotion was aroused... which is great for the entirety of humanity." The visibility that platforms like YouTube provide has significantly raised awareness and excitement around the possibilities of robotics, showcasing what we can achieve.

Optimus robot

> Admiring Elon's ambition, I think it's hard not to see his ventures like Tesla and SpaceX as groundbreaking—and maybe, just maybe, Optimus was inspired a bit by what we did with Atlas. Yet, while ambitious, I feel he's got some way to go in the humanoid robot space, which isn't quite where Atlas is yet.

> At the AI Institute, we've got a 'robot playground' of sorts, with a diverse array of systems like Spots and ANYmal robots. We're ambitiously building our own stuff, but initially, we bought off-the-shelf robots to get things off the ground quickly.

> On competition, I've never been much of a commercial guy; at Boston Dynamics, it was about innovation, not competition. But in the cognitive realm, the battle is fiercer. As for making robots cheaper, there's still a long journey ahead, and I'm optimistic costs will drop significantly with mass production and engineering innovations.

Future of robotics

> Intelligence comes in many forms, and while there are areas where computers may surpass us, I find it puzzling that we assume a smarter machine would pose a threat. There's immense potential for collaboration with these systems if we align their objectives with ours, rather than viewing them as adversarial.

> The Hawaiian shirt I've adopted has become a personal symbol of my contrarian spirit—defying norms and questioning limitations. When faced with skepticism, I often channel a mindset that asks, "Why not?" It's about embracing challenges and pushing the boundaries of what we believe is possible.

Advice for young people

> One key insight I want to share is that life and career choices aren't as limited as many people think. I used to tell my students that if there were no constraints—be it resources or skills—they should envision what they really want to do. I truly believe that “the opportunities really are limitless,” and we must be willing to explore those possibilities, even if it takes years of persistent pursuit to reach them.

> Another profound realization I've come to is that the journey of learning, whether it's robots struggling to walk or ourselves tackling new challenges, is what makes our experiences so meaningful. There’s a unique charm in seeing something “suck” at a task and gradually improve. This process not only reflects our own struggles but also serves as a reminder of how special humans really are in their capabilities.