The Evolution of Robotic Utility: Remote Control for Boston Dynamics’ Spot

Observing Spot, the quadruped robot from Boston Dynamics, navigating a practical environment like an industrial facility presents a curious contrast. Despite the captivating aesthetic displays of its capabilities showcased in viral videos, there's a certain ordinariness in witnessing it perform routine, often unglamorous, and potentially hazardous tasks – precisely the kind roboticists champion.

Over six months have passed since Spot became commercially available, with more than 400 units sold according to Boston Dynamics. During this period, companies have been implementing this advanced machinery in surprisingly mundane scenarios. Recently, a remote piloting experience of the robot was conducted, accessible from a standard desktop computer.

Introducing Scout and the Enterprise Edition

This week, the robotics innovator, now owned by Hyundai, is launching Scout, a browser-based interface designed for remote robot control. This release coincides with the introduction of a self-charging Enterprise edition of Spot, alongside the previously announced Spot Arm.

All of these new offerings are available for purchase starting today through the Boston Dynamics website, utilizing a quote-based pricing system. Importantly, Scout will maintain compatibility with all existing Spot robot versions.

Given Spot’s operational duration of approximately 90 minutes on a single charge, the company recommends pairing it with the self-docking Enterprise model for continuous monitoring applications.

Prior experience driving Spot in person reveals a moderate learning curve, mirroring previous interactions. Boston Dynamics anticipates approximately 15 minutes will be required for full proficiency. However, within just a few minutes, the robot was successfully navigated up and down a staircase at Boston Dynamics headquarters.

The robot is equipped with numerous cameras and sensors, costing $75,000, to mitigate the risk of operational errors.

Remote Control Features and Functionality

The system supports Bluetooth gaming controllers, although the demonstration utilized a keyboard interface. A standard WASD control scheme provides familiar movement options for PC gamers. Arrow keys facilitate switching between four integrated cameras, offering comprehensive situational awareness.

Additional viewing modes are available, including a terrain view that presents a rendered, top-down perspective of the robot and its surroundings. This is particularly useful for identifying immediate obstacles. A picture-in-picture mode allows for simultaneous viewing of multiple camera angles.

A “click to go” feature simplifies navigation. By selecting a point on the ground, Spot autonomously navigates to that location. This functionality is especially beneficial in scenarios with limited bandwidth, such as deployment on remote infrastructure like oil rigs.

Zack Jackowski, Spot’s Chief Engineer, shared with TechCrunch, “[W]e have a power plant customer who had a possible equipment failure. They were able to use the robot to repeatedly inspect something that, if it failed could have been dangerous to a human inspector.” This allowed them to avoid a costly shutdown by remotely assessing the situation.

Enhanced Capabilities and Future Development

A dedicated “stair mode” optimizes the robot for ascending and descending staircases. While manual activation is currently required, Spot can also navigate stairs in standard mode. Current functionality is limited to line-of-sight operation.

Jackowski added, “We have all sorts of crazy plans to extend that to building scale, but the first thing we want to get out there is line of vision.”

The Enterprise edition features a new docking connector, an upgraded CPU, and improved wireless connectivity. It is available either bundled with the docking station or as a standalone unit.

Unfortunately, the new arm was not available for testing. However, Jackowski explained its functionality: “you issue the arm commands, like ‘move your hand here’ or ‘pick up this object’ or ‘turn this valve,’ and the robot’s actually smart enough to figure out, ‘hey, if I’m going to turn that valve, I need to stand over here, I need to shift my weight like this, I need to figure out how to keep the right part of my wrist limp to accommodate how that valve moves.’ ”