Autonomous Systems (AS) are set to disrupt traditional industries, creating a global robotics opportunity that, according to analysts at Tractica, could be worth up to $151 billion by 2020. Assistive technologies such as autonomous robots, self-driving vehicles and other specialized systems can be designed to extend human productivity, build significant business value and create exciting opportunities for developers.
The engineering requirements for AS development — encompassing sensors, controllers, software and communications — are much the same as exist in embedded and mobile computing today. But given the uncontrolled environments in which these systems operate, there are also significant challenges in developing decision making and learning algorithms that realize the intelligence in Autonomous Systems.
To succeed, developers should take into consideration how and where these systems will be deployed, and use this knowledge to select the best hardware platform for optimal performance.
Insight-driven development gives leaders the edge
While major corporations are already making headlines with autonomous vehicle prototypes, they’re also investing heavily in understanding the use cases, opportunities and challenges of Autonomous Systems.
Amazon, for example, hasreportedly established a 12-person team to figure out how to leverage autonomous vehicles, and is hosting supply chain industry events to generate insights into the social, environmental and technical challenges faced in deploying AS in the future. It’s the kind of creative, antidisciplinary design approach we discuss in our previous QDN post on Hardware-Software Convergence. While the technology powering AS continues to be refined, its success is contingent on the quality of insight supporting its development.
Hardware and software working together to realize true autonomy
With opportunities to deploy these systems in the warehouse, in the air, on the road or elsewhere, the value created in AS lies in an orchestration of hardware and software. It’s critical to select a robust and versatile platform to support the varied processing demands of AS, alongside powerful development tools and libraries.
Together, the Qualcomm Snapdragon mobile platform, supporting true heterogeneous computing, alongside the Snapdragon Neural Processing Engine SDK and Snapdragon Profiler, are designed to improve the performance of existing AS and machine learning applications. For example, the Google TensorFlow machine learning framework is now optimized for the Qualcomm Hexagon 682 DSP, and is engineered to allow apps to run with efficiency on mobile and embedded devices.
In April, we announced a collaboration with Facebook to support the optimization of Caffe2, Facebook’s open source deep-learning framework and the Snapdragon neural processing engine (NPE) framework. One of the benefits of Snapdragon and the NPE is that a developer can target individual heterogeneous compute cores within Snapdragon for optimal performance. The NPE includes runtime software, libraries, APIs, offline model conversion tools, debugging and benchmarking tools, sample code, and documentation.
True autonomy, however, will be realized through the software running on these platforms. Developers can model environments to anticipate the kind of actions their AS will perform in the field, but more critical is building the learning algorithms that allow the AS to respond appropriately to changing or challenging environments. It’s a significant computational feat requiring rapid processing of multiple sensory inputs, nimble decision making and robust operational execution.
Communications underpin the entire effort. Within the system itself, of course, but also between systems, between the AS and the cloud or the AS to a human controller. They’re all integral to the performance and usability of the system. While robots deployed on premise may be able to get by on high-speed wireless mesh networks, it’s likely to be a combination of Wi-Fi, 4G and new 5G cellular networks to deliver the flexibility and redundancy that can help operate in an uncontrolled environment.
Autonomous Systems: An exciting opportunity for developers ready to innovate
There’s little doubt that our lives and those of our children will be significantly impacted through the development of Autonomous Systems. Whether through substitution or augmentation, our business and personal productivity looks to be greatly enhanced through artificial intelligence — and Autonomous Systems are some of the most exciting manifestations of these changes. While we’re just scratching the surface of what’s possible today, AS offers huge opportunities for developers ready to take on the challenge.
Stay tuned as we dive into the role of AI and the Snapdragon mobile platform in evolving Autonomous Systems, mobile/embedded computing, IoT, and XR.