SLAMcore Blog

Optimized Portfolio

Posted by Owen Nicholson on Mar 2, 2021 5:19:18 PM
Owen Nicholson

Optimized Portfolio_v2

Any software business working in the complex field of robotics will face a choice. Hardware and software form a mutually interdependent solution; hardware requires software to function, and software must be optimized to make best use of the hardware available. There is no single hardware configuration that is right, or will even function, across multiple platforms and applications, so software developers must choose how and with what hardware to work. Most software vendors go one of two directions. They either opt for a consultancy approach – offering to optimize for any hardware configuration, but charging a fee for each customization, or plump for one configuration and effectively re-sell the hardware that supports their solution.

Both approaches have pros and cons, but SLAMcore has chosen a different route. Whilst the consultancy model offers the broadest possible market, it is not scalable, and may shut out many smaller businesses. Consultancy is very hard to scale as every new project involves complex reconfiguration for every sensor, processor and numerous other factors. It can quickly become expensive and time consuming, slowing down innovation cycles and denting commercial viability. Smaller businesses may simply not be able to afford the fees linked to this tailoring of solutions.

Focusing on a single hardware configuration can provide a fast and potentially cost-effective solution – but only to a very narrow niche. Every different type of robot will have different combinations, orientations and placement of sensors determined by a myriad of factors including weight, power, cost and ultimately the application for which the robot is intended. Tying software to specific hardware solutions massively constrains these decisions and will lead to undifferentiated solutions at best, and most likely, less effective robots.

A third way is to be ‘hardware-agnostic’. It is possible to create algorithms that will run on any processor and with any sensor. But I suggest that this means everything has to meet the lowest common denominator creating software so watered-down as to be all but useless.

SLAMcore is pioneering a different route which we are calling our hardware portfolio approach. We will support a small but carefully selected range of sensors and processors and optimize our software to work perfectly with each. Our work with Intel’s RealSense Depth Camera D435i is one example. Working directly with Intel we have optimized our first public SDK to maximize the benefits from their stereo cameras, depth camera and IMU combination. This optimization allows immediate testing of the system capabilities and enabling tuning to match particular use cases.

By creating these tight integrations with some of the leading, and most widely used hardware elements SLAMcore can deliver the best of both worlds. Award-winning Spatial Intelligence algorithms optimized for widely used key hardware elements.   Developers can select the right sensor and processor configurations from a growing list of partners and create combinations that work for their specific robotic application.  In addition, once product testing has been completed the software can also be ported across to run on custom sensors if required to hit a lower commercial price point.   

As well as Intel we are already working with several other leading processor and sensor developers and will soon announce further integrations. Our plan is to always be open and to work with new and leading hardware solutions as they emerge. Hardware partners benefit from the years’ of specialist SLAM leadership that we can bring, and our customers benefit from algorithms that work out of the box with some of the best hardware around. 

It's our mission at SLAMcore to support the robotics industry by making quality Spatial Intelligence available to all. Our approach enables us to scale quickly whilst still delivering the best SLAM solutions around and I believe this combination will lift the whole industry to deliver better robots faster. 



Topics: RobotsForGood, Robotics, VisualSLAM, RaaS, Intel RealSense, Spatial Intelligence