In 2007, then Microsoft CEO Bill Gates wrote in Scientific American that he expected a robot in every home. If that is to become a reality, robots will have to navigate in real time the highly individualized terrain of each home, avoiding the numerous obstacles that frequently change locations. To do so, the robots will need a robust simultaneous localization and mapping (SLAM) system.

SLAM-An Introduction
A SLAM system allows a robot to quickly-ideally, in real time-construct an internal map of any unknown environment, or update an internal map of a known environment, in which it finds itself. If this is not difficult enough, the robot will also need to continually update its own location as it moves within that environment. In short, the robot must rely on a SLAM system to answer two basic questions:

• What does this space look like?
• What is my location and orientation in this space?

Further compounding the challenge, the robot must act quickly and provide itself with as much rich detail as possible.

To accomplish their work, SLAM systems must collect data from the robot’s own sensors and any other sources to produce what amounts to an internal GPS for its own use. Sensors can include cameras, lasers, auditory sensors, or some type of preinstalled or wire-guided system. Often the SLAM is called to collect this data without the aid of light. Mapping enables the robot to use sensor input to create a virtual model of its environment.

Introducing Karto 2.0
In June 2010, SRI International, an independent research and technology development organization that was originally part of Stanford University, announced Karto 2.0. SRI describes Karto 2.0 as the highest performance, full-featured SLAM system available. Sold commercially, Karto is an advanced development project within the AI Center at SRI International.

Accompanying Karto 2.0 is the Karto SDK, which, according to the company, enables developers of mobile robotics products to integrate navigation and mapping intelligence into their designs. The Karto SDK leverages three decades of research by SRI International to provide high-accuracy mapping, navigation, and exploration functionality across a broad range of mobile robotics platforms.

Specifically, Karto 2.0 introduces several new features, including:

• Dual licensing
• GPS integration
• Custom data storage
• Manual loop closure
• Support for large buildings (over 28 acres)

Characterized by the company as a software-only solution, the Karto SDK provides robotics product developers the flexibility to work with the widest range of mobile robot platforms, simulation environments, operating systems, and robotics middleware.

The dual licensing of Karto adds to its flexibility. Developers can opt for the free open-source version, in which enhancements they make to the system will be contributed back to the community, or the commercial version, for developers working on products they intend to market. Along with the open-source version is the Karto open libraries, providing developers with downloadable versions of source code they can modify for their own use. Support services, if desired, can be purchased separately.

The commercial version of Karto, described as SLAM-plus, provides GPS integration, through which localization can be reported in GPS coordinates, and user-stored data. In addition, support for multithreading and multicore optimization enables the commercial version of Karto 2.0 to run three times faster than the open-source version.

Both the commercial and open-source versions of Karto provide a comprehensive software solution incorporating advanced algorithms for map generation, autonomous navigation, and exploration, according to the company. Along with the algorithms, Karto includes a well-defined, easy-to-use application programming interface (API) that developers can use to integrate Karto’s SLAM capabilities into their mobile products.

The Karto Market
Karto has been adopted by robotics companies such as Robosoft, commercial product manufacturers like Segway, and government research organizations, including the Idaho National Laboratory. Karto-enabled products show up frequently in hospitals, and are increasingly being used in unfriendly, even hostile environments, such as border patrol or search and rescue.

SRI International (Karto Robotics) insists that it has no direct competition. According to Karto’s developers, their biggest competitors are software developers intent on building their own SLAM solutions. Other navigation competition comes from camera-based systems, or solutions that use prepositioned sensors or wire guides.

By comparison, Karto works by using a laser rangefinder. It points the laser beam at an object, gets a reflection, takes the measurement, and feeds the resultant data into its SLAM algorithm. Karto’s laser-based approach allows systems to function in light-deprived environments and still sense their surroundings very quickly. The goal is to enable customers to deploy robots in environments they have not previously seen. Using Karto, robots can map a new environment in real time, as quickly as the systems can move.

Karto Technology and Business Model
The business case for Karto revolves around time to market. By using Karto, developers of mobile robotics products avoid the time and effort it takes to develop their own SLAM capabilities. In addition, Karto is field-proven, and the 2.0 SDK provides developers everything they need to build robust navigation capabilities quickly.

Karto is surprisingly lightweight and energy efficient. It uses an Atom 1.6GHz processor with 1GB of RAM. The latest version actually delivers more capabilities then its predecessor while consuming fewer system resources.

The economics of Karto are best understood in the context of the cost of the robotic device with which it is used. For example, the cost of a laser rangefinder can run from $1,500 to $6,000, while the robots themselves may cost $15,000 to $25,000. The Karto software, which costs $5,000 for the commercial license, is comparatively low-priced.