Boston Dynamics’ brand new all-electric robot, Atlas, has attracted worldwide attention, even astonishing tech magnate Elon Musk. The dexterity of Atlas’ joints, almost akin to monsters in movies, showcases its remarkable potential. With the fall of hydraulic systems, the era of electric robots dawns, and the trend of humanoid robots is set to climax this year, undoubtedly bringing new focal points for investors from all walks of life.
Boston Dynamics’ Atlas officially bids farewell to the hydraulic age and announces the arrival of a new era. Competition in the commercial humanoid robot sector is intensifying, and Boston Dynamics is evidently ready to compete with other giants. Their newly released video surpassed a million views shortly after release, demonstrating its widespread attention.
In the video, Atlas starts with a calm lying position, showing incredible bodily flexibility. After standing up with breathtaking movements, it turns its head to reveal a circular screen with a ring light, silently announcing a technological breakthrough. People commend its extraordinary range of motion and progress, while also being shocked by the robot’s capabilities that almost surpass those of humans.
Boston Dynamics’ decision to replace hydraulic systems with electric drive is widely regarded as an inevitable choice in technological development. The birth of the new electric Atlas not only continues Boston Dynamics’ pursuit of humanoid robots but also heralds the arrival of an even more advanced technological era. Boston Dynamics is beginning to define the Atlas humanoid robot as a product, indicating the embryonic form of its commercialization and the promising prospects of its future widespread applications.
The launch of the all-electric Atlas is indeed exciting, its new design adopting batteries and electric actuators, providing unmatched strength and flexibility. Its appearance is a breakthrough in technology and represents a significant change in appearance. Compared to the previous, bulky hydraulic Atlas clad in armor plating, the new generation Atlas offers a more approachable and advanced image.
The newly-designed Atlas robot displays a more refined silhouette, with a more flexible mechanical structure, and no external wires exposed. This to a certain extent reflects Boston Dynamics’ response to public concerns. Compared to the initial impression left by the original Atlas, the Figure 01, and Tesla’s Optimus, the changes in the new Atlas design seem to be pacifying the public’s fear and suspicion of a large-scale robot uprising.
In the latest trailer, what draws the most attention is Atlas’ fluid and smooth movements. This design makes us wonder: do humanoid robots have to completely mimic the human form in appearance? Investors have said that despite billions of years of evolution, humans have not become perfect mechanical beings. If we are to create bionic machines, why not design a machine that can perform tasks that humans cannot? In light of this, Boston Dynamics installed a set of custom high-power actuators in the joint areas of the Atlas robot. This innovative design is highly flexible, greatly expanding Atlas’ range of motion, as if encapsulating the power of a top athlete within a compact body.
The shocking inverted leg movement of Atlas that startled many viewers was not just for show, but also highlighted its practical application value in an industrial environment. When most industrial robots malfunction, they often require manual intervention to resume work, but if robots can reset themselves autonomously, then work efficiency would be greatly improved. Atlas’s ability to pivot quickly is equally significant, as robots frequently need to shuttle between shelves and conveyors during work. If the time for each shuttle can be saved, then the cumulative time saved can be considerable. When moving in tight spaces, a smaller turning radius is also more in line with current workflows and spatial constraints.
The modern version of Atlas, unlike traditional designs, features hydraulic Atlas’s distinctive three-fingered hand, rather than the common four-fingered design found on the market. To manufacture a robot capable of performing a diverse array of tasks, ensuring the reliability and durability of its actuators when contacting external objects is crucial. This is why Boston Dynamics chose to reduce the number of Atlas’s fingers to fewer than five. Reportedly, they are continuously improving this design to adapt to the flexible grasping of objects of different shapes and are integrating advanced sensing systems to precisely detect objects.
Another feature of Atlas is its large and round head design, very much reminiscent of a “makeup mirror.” This is an element consciously created by the design team, aiming to break the stereotypical image of humans and robots. The company’s CEO revealed that this design was inspired by the small desk lamp image from Pixar Animation Studios. “We want it to look friendly and open, not only can the head be used as a display, but its shape itself also exudes friendly intent, which is crucial for future human-robot interaction.”
In summary, compared to the robotic prototypes people imagine, the electric version of Atlas appears more advanced and eye-catching. It has omni-directional joints, 360-degree audio and video sensing capabilities, and breaks away from the traditional framework where design must adhere to the “human species” appearance. This is a significant sign of humanity moving towards higher levels of creativity and capability.
Jim Fan, the leading figure of NVIDIA’s AGI Moon Landing Project GR00T, proposed a thought-provoking viewpoint: “We are obsessed with pursuing ‘human-level’ robot technology, yet overlook that this is just a man-made limit. Since technological innovation is endless, why not aspire to create a new species that surpasses humans from the start?”
As for the commercial humanoid robot field, the tumultuous changes over the past decade are particularly noteworthy. Boston Dynamics’ announcement of the retirement of its iconic robot—the hydraulic version of Atlas, has been interpreted as a firm signal of the company’s entry into the commercial field. “Do they want to replicate Optimus Prime?” Although many believe that Boston Dynamics’ technology is far ahead of Tesla’s, the high cost of heuristic methods makes Atlas’s retirement reasonable. Undoubtedly, from a commercial standpoint, Tesla seems to have the upper hand.
Some opinions even compare Atlas with Tesla, which is like comparing Ferrari with Ford Model T. Though both are cars, their product positioning and functionality are fundamentally different. Since Boston Dynamics was acquired by Hyundai Motor Group in 2021, it has clearly accelerated its pace toward commercialization. Emerging companies in the market, such as Agility, Figure, 1X, and Apptronik, have stirred up the humanoid robot market. Hence, Boston Dynamics has also started to seriously consider the feasibility of commercializing the humanoid robot market.
In fact, in terms of current technology, Boston Dynamics indeed occupies a leading position in humanoid robot technology. Meanwhile, hydraulic systems have been timely phased out of history. Regarding the retirement performance of Atlas yesterday, especially that fall where hydraulic oil was sprayed, many spectators described their heartache with the phrase “it hurts to watch”.
Robert Playter, the CEO of Boston Dynamics, has been immersed in the field of robotics for 15 years. IEEE SPECTRUM secured an in-depth interview with him:
Question: How is the current development? Why has it taken so long?
Answer: The all-electric Atlas represents the culmination of our team’s nearly 15 years of effort in developing humanoid robots. We have always believed that it’s possible to build a humanoid robot that can handle a variety of tasks. This new model is capable of manipulating or moving heavy objects with complex shapes with both hands, which is far more complex than our previous robots. Additionally, the electric Atlas was developed in parallel with the traditional hydraulic Atlas. Our previously released videos have already proven these technologies are feasible. Our current goal is to design and build a new generation of robots that surpass any existing machines and plan to introduce them to the market as a product in the future, with a primary focus on industrial and logistics applications.
Question: In the development of the robot dog Spot, it seems that Boston Dynamics didn’t have a predefined application scenario?
Answer: In fact, for Spot, we chose to develop and bring it to the market directly, allowing users to discover its potential applications on their own.
When developing the strategy for Atlas, there were some different considerations compared to previous products. Currently, development for specific use cases has become a premise, as evidenced by the introduction of the Stretch product aimed at clear market targets. Atlas’s functions are not limited to single tasks; its vision is to enable a wider range of applications.
When it comes to the vision of Atlas, the goal is to become a feature-rich “multipurpose” robot. There are few robots on the market that can extend the ability to perform a single task to more complex work. Atlas will carry such expectations and turn its focus to how to make breakthroughs in performing diversified and complex tasks.
Considering Atlas as a product in the future, rather than just a research platform, will bring about a series of changes. Research over the past 10 to 15 years, in ensuring robot mobility and object picking capabilities—including dynamic balance and mobility—has been crucial. However, there is still room for improvement in terms of universality, which will depend on the advancement of key technologies such as artificial intelligence. Extensive prototyping and iteration are needed before robots can be mass-produced and delivered to customers.
Regarding whether the “hydraulic” system is the best way to achieve Atlas’s dynamic motion (once a common view), there is now a new understanding. We’ve gone down a similar path with the development of the Spot robot, subsequently developing lightweight electric motors with excellent responsiveness and strength. Then, we designed a new type of “compact actuator” for the electric Atlas, integrating “professional athlete” level strength into a smaller device. The results show that the new Atlas surpasses humans in both strength and range of motion, and the performance of the electric version is better than the hydraulic version.
The video used to demonstrate Atlas’s range of motion has raised some objections, as some viewers may feel uncomfortable. However, these highly flexible actuators give the robot many unique movements and greatly improve its work efficiency. Such flexibility offers endless possibilities for innovation and the development of new functions, something we are excited about and have decided to share with global audiences.
Regarding concerns that robots need to move in a familiar and predictable manner so that people around them do not feel uncomfortable or in danger, we are already taking this into consideration to ensure that the robot’s design is both efficient and friendly.
Playter: Indeed, limiting robots to merely imitate human movements is not enough. If robots can be designed to be more powerful or flexible, then they will be able to perform tasks that humans cannot or are unwilling to handle. For example, the joints of the first generation of Stretch robots could not achieve 180-degree rotation, making its movements to pick up and place items quite sluggish. However, with the addition of a rotating joint, Stretch’s movements have become swifter and more efficient. As a result, people have gradually become accustomed to this new way of moving.
Spring: The headless design makes the Atlas robot vastly different from its predecessors, and the new head design not only allows the robot to express intent but also integrates a communication lighting system. Initially, we tried to design a more human-like appearance, but that design always brought a sense of unease to people, as if it came from some dystopian sci-fi scenario. We ultimately deliberately chose a round head design to reduce its human likeness and convey a friendlier vibe. Moreover, this design also drew inspiration from the well-known Pixar Animation Studios’ Luxo Jr. lamp.
Spring: Years of experience have been crucial for Atlas to become a product. It’s not enough to have a range of interesting technologies to guarantee success. A product must be targeted towards specific applications and provide real value. We can only say that we’re officially in business when we’re able to sell multiple units of a robot to a customer. This can’t be achieved without sufficient reliability, customer service, and system integration. Years ago, we started bringing Spot to market, which in fact marked the start of a strategic transformation for the company. For this transformation, we had to fully delve into previously uncharted territories, such as manufacturing, servicing, assessing robot quality and reliability, and establishing systems and tools for ongoing improvement. At the same time, we needed to produce at a reasonable cost, sell at a sensible price, and ultimately make a profit. Such a change typically takes two to three years to complete. So, those startups with only prototypes may not have even begun this trek. At present, we are not in a rush to announce any technological breakthroughs, nor do we easily show off our ambition unless a viable product development path has been confirmed. I believe that’s the key to success.
Spring: We are about to showcase more control technologies on the upgraded version of Atlas that have already been tested on the previous model. We expect that as early as next year the new robot will be able to enter modern automotive factories for technological verification tests.
Spring: Regarding the new Atlas, we believe what will interest people the most will be its lightweight, small size, high flexibility, and powerful functions. Once people see its performance, they are sure to exclaim: “Wow, it can actually be done this way!” For years, other companies have been using us as a benchmark, and now that standard is being raised again.
I believe this will ultimately drive the entire industry forward.
