Spot (Source Boston Dynamics)

See SPOT Do More Than Run — Boston Dynamics Mobile Robots

Vincent T.
0xMachina
Published in
6 min readFeb 8, 2021

--

If robots are going to take over the world, it is going to begin with Spot. That’s the now well known robot built by Boston Dynamics. Spot has been evolving, with new capabilities and advanced features as a result of research and development. What makes Spot different from other robots is that is very mobile, giving it abilities that one can consider remarkable. Alongside Spot are other robots from Boston Dynamics like BigDog, Atlas and Handle. While these robots definitely do not look anything like the Terminator or Autobots in movies, it is what robotic applications are like in the real world.

Robots are being used throughout the world in a variety of applications. That includes mostly industrial (e.g. automobile manufacturing) and commercial (e.g. warehouse shipping, factory packing) applications. The Boston Dynamics robots are developed to face challenges. As they explain it, the robots purpose are to provide solutions that use “advanced mobility, dexterity and intelligence.” These are robots that perform tasks, either autonomously or by programmed control, using human-like (or animal-like) movements.

Spot is a type of mobile robot with 4 legs, giving it a semblance to a dog (Source Boston Dynamics)

Spot

Spot falls under the class of mobile robots, that are used for tasks like surveying terrain and gathering real-time data. What makes Spot truly familiar is its design. It is very much like man’s best friend, the dog. It does not resemble the robot you would see on the industrial assembly line or warehouse floor. This is because Spot has four legs, just like a real dog. That design was motivated by functionality, to perform tasks that require a certain type of mobility that is more natural. Like a dog, Spot can walk or run (though not as fast as a real dog) much better with legs compared to having wheels or tracks for mobility.

Spot uses its legs to climb stairs, overcome obstacles and to travel from one point to another. With these capabilities, Spot would be ideal for tasks that would require more natural movement. An example of this would be to navigate through rugged terrain, which tend to be uneven and full of obstacles at every direction. If Spot had wheels like a rover, it would have a much harder time moving around that type of environment. This is the advantage of having robots with legs.

There are two versions of Spot. There is the Spot Explorer and the Spot Enterprise. The former is a more basic type of robot, while the latter has more advanced capabilities. The robots use rechargeable battery for power, with the Enterprise version having its own power dock. They are capable of autonomous operation with network support using Wi-Fi (2.4 GHz on Explorer and dual-band for Enterprise).

The models of Spot, Explorer and Enterprise versions (Source Boston Dynamics)

A more recent development in Spot, is the addition of a mechanical arm. While Spot was originally just a mobile robot, by adding an arm it now has the ability to manipulate objects. This upgrades the features of what Spot is capable of doing. Now that it has an arm, it can interact in a new way with the environment around it. This adds new applications to its array of features.

A Spot robot with an arm (Source Boston Dynamics)

Components

The key components for robots like Spot are not just the electro-mechanical systems for motion. Spot uses on-board AI software (developed by Vinsa) that uses deep learning techniques for autonomous operation. It uses this with a system of 5 stereo cameras equipped with global shutter, and embedded greyscale image sensors There are 2 in the front (1 on each side) and 1 on the back. The software works with a vision system to gather data and to help the robot explore and navigate.

The system uses Simultaneous Localization And Mapping (SLAM) that provides the robot the autonomous navigation features. A GPU is also utilized by the vision system to render and analyze images from the cameras. The core engine of the software uses TensorFlow deep neural networks that were built on deep learning models. This helps to improve performance, with experts becoming part of the feedback loop.

An example SLAM diagram (Source SLAMcore)

SLAM uses algorithms for robotic mapping methods that give Spot information about its position relative to the environment .There is always the unknown factor when it comes to environment. Spot needs to be aware of its environment in order to arrive at the best decision. If Spot makes the wrong move, it can be costly, which is why it uses learning techniques to better understand its surroundings.

Applications

If you look at the customers Boston Dynamics works with, they are mostly in industry or academia. That is because robots like Spot are used mainly for specialized tasks. Spot can help these companies tremendously when it comes to gathering data and taking the place of a human when it comes to more hazardous conditions. Perhaps that is one of the more important considerations for using robots.

Spot can be deployed for performing inspection by construction companies, to check the stability and progress of a project. Spot could be fitted with cameras and sensors to gather data at the construction site. It then relays the data back to a command center which can apply analytics to create information for the site inspectors. It could take longer for a human to perform that same task, thus giving Spot the advantage.

Spot can also be deployed to areas that would otherwise be dangerous for humans. Hazardous work environments like chemical plants, oil facilities and gas pipes can make use of robots. Spot can be fitted with sensors to detect leaks that would otherwise be toxic to a human worker. If the area is radioactive, it can gather data without the risk of sending a human who can get exposure to the fallout. Spot can also provide eyes on location. It can be deployed into mining tunnels to check on the safety of the location before actual miners are allowed to go in.

One way Spot can be deployed is on site for surveys and inspections at a facility (Source Boston Dynamics)

Most of these applications don’t require performance of manual labor. Instead, Spot is more for gathering data and providing information remotely. Data like temperature, chemical emissions, photographs and measurements. It can be deployed on site first to perform surveys, as well as to check on the safety of an environment. From there other functions can be added. With the use of an arm, it can pick up objects and analyze it in real time. Spot can even be programmed to perform chores around a site, like cleaning floors or removing obstacles.

Another application for Spot would be for exploring and gathering data in hostile environments like the surface of Mars. Rovers have been sent in the past to perform tasks, but there is always the problem of mobility. Perhaps Spot type robots can move around better on rocky and obstacle filled environments.

Synopsis

The use of robots like Spot, expand on the abilities of legacy robots. They have more mobility, autonomy and were designed with new challenges to overcome. When it comes to gathering data for providing information, Spot can be deployed for inspections, surveys and reconnaissance on site. They are more versatile in a sense that they can perform tasks better or that could not be performed before by automated means. Developers can also customize Spot according to their own requirements and specifications using an SDK (Software Development Kit).

The enhanced capabilities truly make Spot not just a one-task machine, but a multi-purpose unit. While they may replace humans in some tasks, it can actually be for the better in terms of safety. Spot can be deployed to prevent risk to human workers when needed. That is yet another benefit we can realize from these new class of robots. This can truly change the idea of what robots are capable of doing.

--

--

Vincent T.
0xMachina

Blockchain, AI, DevOps, Cybersecurity, Software Development, Engineering, Photography, Technology