Overview: Boston Dynamics stands alone as the leading developer of highly dynamic robots possessing lifelike mobility. Founded in 1992 as a spinoff from MIT, they have pushed boundaries by tackling the most difficult problems in bipedal and quadrupedal motion. Sustained funding from military sources like DARPA enabled long timelines to achieve milestone breakthroughs in robot locomotion and dexterity. While initially focused on research, recent robots reveal a pivot towards practical commercial applications – especially in logistics. However, concerns linger regarding advanced autonomous robots and their potential risks beyond strictly controlled environments. As both a tech industry analyst and robotics enthusiast, I closely track Boston Dynamics‘ remarkable progress advancing the state of the art while also commercializing real world robotic solutions.
The Origins Story
Professor Marc Raibert founded Boston Dynamics in 1992 based on years of prior research into dynamically balanced robots. As a former MIT professor and founder of the Leg Lab in 1980, Raibert recognized bipedal robotic locomotion as an enormously difficult but potentially game changing challenge. His key insight was that balance and stability are integrated problems that cannot be decomposed into separate simpler blocks. That led to foundational work on systems balancing themselves using sensors, computation and feedback loops.
Early funding came from military sources aimed at improving robotic navigation across rugged real world terrain. Tackling these extremely ambitious problems required long timelines unconstrained by commercial milestones. This suited Raibert, allowing him to slowly assemble an expert team to iterate on simulations before attempting physical platforms.
The expertise built across locomotion fundamentals, mechanical design, sensory-motor control theory and navigation algorithms positioned them to overcome barriers that had halted other robotics labs. Their lab was small but stacked with talent focused maniacally on mobility breakthroughs.
Timeline of Key Events
| Year | Event |
|---|---|
| 1992 | Boston Dynamics founded by Marc Raibert |
| 2005 | BigDog quadruped robot unveiled |
| 2008 | PETMAN humanoid robot for testing hazmat suits |
| 2012 | Google acquires Boston Dynamics |
| 2013 | Atlas humanoid robot created |
| 2017 | Sold to Softbank |
| 2020 | Hyundai acquires company |
| 2021 | Stretch warehouse robot released |
Early Quadruped & Biped Robots
Marc Raibert built upon his prior research into dynamically balanced legged robots. Traditional robots use static stability – carefully controlling the center of gravity over the support legs. This constrains range of motion. Animals achieve much greater mobility through dynamic stability – constantly sensing balance and making micro-corrections to remain balanced even as legs swing into new positions.
BigDog demonstrated this principle using precision leg control to hike over ice, rocks and slopes. But as an early prototype it was tethered to offboard power and computers while making loud hydraulics noises.
PETMAN focused capabilities towards mimicking human gait, posture and balance. This humanoid form enabled testing hazmat suits in realistic dynamic conditions. Although tethered, it markedly expanded prior bipedal movement capabilities.
RHex took a different form – a mini tank propelled by six interconnected motorized legs. This created a robust, scrambling gait able to cover rough terrain without any balance calculations. Unique Savings in sensing-computation led to high reliability and effectiveness.
LS3 (Legged Squad Support System) aimed to improve on BigDog with a quieter, stronger and faster quadruped robotic mule. It demonstrated carrying 400 pound loads faster over tougher terrain. But power and noise issues remained around hydraulics.
Across these platforms, we see creative exploration into different forms of mobility – quadruped, biped and hybrid. Each expansion embodies pioneering theory into control, coordination and balance for traversing the real world.
| Platform | Type | Capabilities | Notes |
|---|---|---|---|
| BigDog | Quadruped | Picks way over rocks/ice | Noisy due to hydraulics |
| RHex | Hexapod | Fast scrambling movement | Fixed gait but effective |
| PETMAN | Bipedal | Walking, balance | Tethered power |
| LS3 | Quadruped | Faster, 400lb payload | Still loud |
Atlas – A Watershed Humanoid Robot
The bipedal humanoid Atlas platform demonstrated a watershed evolution from Boston Dynamics‘ prior robots. Traditional humanoid robots like Honda‘s Asimo navigate environments using explicit planning. This pre-maps movements step-by-step to avoid dynamic situations risking falls. Boston Dynamics recognized that the future lay in reactive planning – perceiving the world continuously and adapting accordingly using onboard computation.
This key insight on reactive vs planned movement deeply influenced Atlas‘ design. Balancing on two legs is inherently unstable unlike four, six or wheeled platforms. But its upright six-axis form factor enabled dexterous human-like movement previously unattainable. This brought Atlas‘ capabilities closer to a human while retaining versatility to use wheels or limbs as needed for situational optimization.
Introduced in 2013, early versions of Atlas focused on fundamental capabilities like balancing, walking on uneven terrain and righting itself after falling. But rapid upgrades brought major milestones:
- Dexterous hand-eye coordination picking up objects
- Robust builds with protective casings to handle falls
- Completely untethered, battery powered designs
- Upgraded sensor suites (Lidar, stereo cameras, IMUs) for complex environment mapping
- Navigation integration enabling analysis of terrain geometry to plan foot placements
- Expanded movement repertoire including ducking, jumping over barriers
- Remarkable demonstrations like parkour circuits and backflip maneuvers.
These videos of Atlas progression demonstrate bipedal movement once only possible in humans:
Atlas 2013
- Slow, tethered, walking on flat ground only
Atlas 2018
- Robust but slow, basic environmental sensing
Atlas 2021
- Untethered, dexterous mobile manipulation
The sleekest 2021 version reveals the template of a true mobile manipulation platform – able to dexterously interact with the physical world much like a human worker.
Ongoing upgrades to Atlas hardware, software and movement capabilities continue expanding possibilities. This platform remains a remarkable concentrator of Boston Dynamics‘ full stack robotics breakthroughs in mechanical design, sensory-motor control, navigation, mapping, planning and manipulation.
| Atlas Version | Key Capabilities |
|---|---|
| 2013 | Tethered, limited flat ground walking |
| 2018 | Fully untethered, basic environmental sensing |
| 2021 | Dexterous mobile manipulation |
Commercialization Begins
Under Softbank ownership, Boston Dynamics began commercializing robots for enterprise applications rather than just demonstration platforms. The quadruped Spot incorporated modular self-righting sensory capabilities proven across prior projects into a compact yet rugged robot. Its versatile inspection, data collection and communications capabilities drew natural demand from fields like utilities, construction and energy companies to automate hazardous site mapping.
Priced at ~$75K, Spot saw early deployments by specialized firms contracting inspection services. But subsequent releases enabled custom payloads and third party apps – opening uses for biohazard cleanup, healthcare delivery in hospitals and thermal screening capabilities. Public safety agencies also employ Spot for inspecting suspicious packages and environment scans to supply visual intelligence without endangering personnel.
In 2021, the company announced Stretch – a mobile warehouse robot designed to automate box handling via an arm and smart gripper. Unlike Spot‘s specialized use cases, Stretch directly targets massive conventional enterprises grappling with surging ecommerce volumes, warehouse labor shortages and unsafe manual material handling processes. With ~50% of warehouse costs tied to labor, its versatility handling hundreds of SKUs, dexterity to manipulate varying box sizes and durability for continuous operation can justify $200-300K price points through immense labor savings.
Analyst projections show the warehouse robotics market growing over 15% annually the next decade to top $30B:
Warehouse Robotics Market Size estimates – Source: Interact Analysis
With Stretch, Boston Dynamics can scale an enterprise playbook perfected through Spot‘s specialized use cases – combining robotics breakthroughs with practical commercial solutions demonstrating clear ROI.
| Platform | Key Capabilities | Applications |
|---|---|---|
| Spot | Mobility modules + payloads | Inspection, Biohazards, Healthcare, Public Safety |
| Stretch | Warehouse mobile manipulation | Box handling, Pallet loading/unloading |
The Road Ahead
Born from locomotion fundamentals, Boston Dynamics continues updating existing robot platforms while expanding into new ultra-dynamic designs.
Wheeled “Handle” robots adapt their advanced mobility and manipulation for logistics box handling roles. Imagined use cases see Handle cooperating with Stretch, motorizing pallet jacks and reducing injury-prone manual efforts.
Digit biped designs reduce Atlas complexity for industrial applications like drill carrying for car assembly plants. Focus on simple key tasks promises commercial impact despite reduced versatility.
And in 2021, Boston Dynamics introduced the concept robot Atlas Enhanced (“Parkour Atlas”). With even more dynamic athletic capabilities, it promises to continue pushing boundaries on freely traversing environments only humans can navigate currently.
Ongoing areas for innovation remain:
- Increasing reliability and durability for at-scale commercial deployments
- Enhancing grasping and manipulation of previously unseen objects
- Broadening situational awareness via panoramic 3D environmental sensing
- Expanding movement repertoires and contact strategies for more contexts
- Augmenting behavioral control and reinforcement learning for more intuitive interactions
Expert Predictions
As both an industry analyst and robotics enthusiast, I see Boston Dynamics uniquely positioned to lead the next 20 years of commercial robot evolution via breakthroughs across various mobility schemas – bipedal, quadruped and hybrid wheeled/legged combinations. They stand apart in their command of dynamic balancing, terrain adaptation and understanding fundamental tradeoffs between mobility vs. manipulation vs. specificity of use cases. Their technical talent and funding history also seem poised to achieve the hard-to-duplicate innovation mix of highly dynamic obstacle traversing mobilized robots cooperating with dexterous manipulators designed for human-level versatility.
Upcoming years may see an expansion into logistics process augmentation applications where their robots handle material flows around semi-structured depot environments while humans operate sophisticated warehouse management software and handle exceptions. I expect continued growth for their specialized inspection offerings while industrial manufacturing and intralogistics robots drive higher volume adoption. However, the next 5 years will still see more narrowly scoped highly technical deployments rather than mass adoption. Beyond 2030 as capabilities and reliability improve via machine learning, we may witness autonomous Boston Dynamic robots penetrating everyday environments alongside humans.
Of course, with such rapid progress comes ethical concerns about oversight should these humanoid machines become completely autonomous. Today’s models focus on perfecting mobility while keeping humans in command via remote operators – but there is recognition that true economics benefits arrive only when machines operate largely independantly without remote operators. As advanced nations debate guidelines for autonomous vehicles on public roads, I expect similar attention on autonomous advanced robots introduced into public spaces. Regulation will likely encourage restrictive use conditions for the coming decade until societal confidence grows via accumulated safety track records.
I hope this analysis offers useful insights on Boston Dynamics – their remarkable history that pioneered concepts proven impossible elsewhere while also commercializing real-world robot assistants ready to unlock enormous human productivity and take on dull, dirty and dangerous jobs many gladly cede to our autonomous mobile manipulator helpers!
