There has been an explosive growth of service robots in the commercial sector as evidenced by the many different use cases for them. For instance, service robots can be used as mobile signages and self-service kiosks. Such usage requires the support of various kinds of workloads associated with those functions. It necessitates the deployment of a capable compute module reference design that can meet the expanded and varied workloads. The challenges, therefore, can be listed as follows:
- Support for advanced navigation and analytics: Currently, modern service robots use computer vision and depth sensors/cameras to complement LiDAR for object recognition and collision avoidance. With the expansion in the kind of usage, the compute module not only needs to provide the computation power for complex navigation and maneuvering, it also needs the headroom to support AI workload, media processing, and such other operations.
- Scalability and time to market: Service robots belong to a very fragmented industry. Robots are available in a range of sizes and capabilities depending on the application. The design and validation cycles required for each use case are extensive and can hamper the product’s time to market. Such delays add to the cost and may not be feasible for small scale deployment.
A scalable compute module design is one that can accommodate the different workloads that a service robot is expected to execute. Such a design eliminates the need for constant redesigning and reduces the product’s time to market.
The Intel® Mobile Smart Kiosk Design has been developed specifically to address these issues in the service robot industry. This paper explores an extensible approach for a service robot compute module reference design to accelerate time to market of service robot designs.