Efficient ROS-Compliant CPU-iGPU Communication on Embedded Platforms

Many modern programmable embedded devices contain CPUs and a GPU that share the same system memory on single die. Such unified architecture (UMA) allows programmers to implement different communication models between CPU integrated (iGPU). Although simpler model guarantees implicit synchronization at cost of performance, more advanced allows, through zero-copy paradigm, explicit data copying iGPU be eliminated with benefit significantly improving performance energy savings. On other hand, robot operating (ROS) has become de-facto reference standard for developing robotic applications. It application re-use easy integration software blocks in complex cyber-physical systems. ROS compliance is strongly required SW portability reuse, it can lead loss elude benefits communication. In this article we present efficient techniques CPU–iGPU by guaranteeing standard. We show how key features each are maintained corresponding overhead involved compliancy.