Behind some remarkable results like those of “Boston Dynamics” are almost 30 years of research in the behavior of mechanical systems and in the study of biomechanics (kinematic and dynamic optimization, trajectory optimization), mathematical optimization for unstructured and dynamic environments (quadratic programming), innovations in the engineering of software and hardware control units, innovations in athletic mobility (sequential composition), innovations in coupling mechanics (e.g., there are 28 miniaturized hydraulic couplings, a miniaturized pump and compact batteries in the Atlas robot), the development of predictive control models and real-time correlation models of information about the surrounding environment with the robot’s movements, etc.

Without deep knowledge of the fundamentals of engineering in the field of mechanics of dynamic systems and adaptive control, such a thing could not have been achieved. For now, Boston Dynamics robots have AI elements only in 3D image processing. The quality of the results in the case of Boston Dynamics lies in the “holistic design”, that is, the hardware, software, and behavior parts are developed simultaneously, interconnected (in harmony) and the physics of the environment is taken into account.

This is in opposition to the traditional approach in which modules made by various manufacturers are “linked” into a robotic architecture. This is true engineering – a holistic one that masters every behavior down to the lowest level. It is about “organic design”, based on massive data associated with the entire behavior of the robot.

The future focus will be on the cognitive intelligence of these robots, on reducing the weight by 10% and on increasing the robustness by 50%, and on new algorithms for optimizing trajectories. In 2020, Hyundai purchased Boston Dynamics from Soft Bank Robotics for $1.2 billion, which bought it from Google X Lab, which acquired it from DARPA (Defense Advanced Research Projects Agency). Boston Dynamics was a research initiative in the science of robotics. Probably in the future, the focus will be on commercial robots (e.g., the “yellow” robot called Spot is sold on the market for $75,000).

The Atlas robot will be a robotic system dedicated to the science of robotics, without a commercial focus, although at some point it could be found in the “law enforcement forces of the future”. Boston Dynamics robots are not yet autonomous, they are remote-controlled (man-in-the-loop), therefore the future focus will be on autonomy. However, even today these robots have SLAM algorithms for autonomous navigation.

But what are other technologies embedded in the Atlas robot?  Some of the key technologies and mathematical models behind the Atlas robot include:

(a) Robotics: Atlas utilizes state-of-the-art robotics technology to enable it to perform a wide range of human-like movements, including walking, running, jumping, and even backflipping. The robot’s design is based on principles of biomechanics, kinematics, and dynamics, allowing it to maintain balance and stability even when moving through challenging terrain.

(b) Machine learning and artificial intelligence: The Atlas robot incorporates machine learning algorithms to enable it to adapt to new environments and perform complex tasks. These algorithms help the robot make real-time decisions about how to move and interact with its surroundings.

(c) Control systems: The Atlas robot is controlled by a sophisticated control system that integrates sensory data from various sources, including cameras, lidar, and IMUs, to generate control commands for the robot’s actuators. The control system is based on mathematical models of the robot’s dynamics and kinematics, which are used to generate accurate predictions of the robot’s behavior.

(d) Computer vision: The Atlas robot is equipped with computer vision systems that allow it to perceive and understand its surroundings. These systems use machine learning algorithms to process images and other data from cameras and other sensors to generate a detailed map of the robot’s environment.

(e) Simulation and modeling: The development of the Atlas robot involved extensive use of simulation and modeling techniques to test and refine the robot’s design. These simulations were used to validate the mathematical models that underpin the robot’s control systems and to optimize the robot’s performance in various scenarios.

The dynamic model behind the athletics of the Atlas robot from Boston Dynamics is based on the principles of physics, specifically rigid-body dynamics. The dynamics of the robot’s movements are modeled as a series of interconnected rigid bodies, each with its own mass, center of mass, and moment of inertia. The movements of these bodies are governed by the laws of physics, including Newton’s laws of motion and the principles of energy and momentum conservation.

To generate the robot’s movements, the control system uses the dynamic model to calculate the required torques and forces that must be applied to each joint in the robot’s body in order to achieve the desired motion. These calculations take into account factors such as the robot’s mass, its center of mass, and its current velocity and acceleration. The control system then sends commands to the robot’s actuators to apply the required torques and forces, causing the robot to move in the desired way.

In addition to the dynamic model, the control system for the Atlas robot also incorporates models of the robot’s kinematics (the geometry of its body and its motion), as well as models of its sensory and motor systems. These models are used in combination to generate precise and accurate control commands that allow the robot to perform a wide range of athletic movements.

Note from the author of this post

It is important to be aware that relying on a single source of information can be risky. While it may be convenient to get all your information from one place, it can limit your understanding of a topic and potentially expose you to biased or inaccurate information. To get a more complete picture, it is recommended to seek out multiple sources of information, including those with differing perspectives. This will help you to form a better opinion based on a wider range of information and increase your chances of identifying any biases or misinformation. In today’s age of information overload, it can be difficult to know which sources are credible and trustworthy. It’s important to evaluate sources critically and consider the potential biases and motivations behind the information being presented. By seeking out multiple sources and critically evaluating the information presented, you can improve your understanding of complex topics and make more informed decisions.


credits: Stelian Brad