Enabling a Connected World with Edge Computing and Sensor Fusion
Advances in sensor technologies and computing capabilities coupled with reduction in prices has led to an exponential increase in the number of “things” connected to the internet. As a result of it, there is an increased adoption of sensor fusion and edge computing systems across industries, from autonomous vehicles to agriculture.
Manufacturing industries would be an early adopter of sensor fusion and edge computing to enable existing IoT implementations in order to bring speed and agility in production workflows in real-time. Smart industrial robots deployed on factory floors could become more responsive and self-sufficient if they are able to process data at the edge. These could facilitate quicker extraction of insight and increased application of machine learning algorithms.
Smart farming could be another area where IoT in combination with sensor fusion and edge computing can bring efficiencies. More often than not, these agricultural farms are located at remote and hostile locations which result in connectivity and bandwidth issues. In such scenarios, edge computing can help smart farms to effectively monitor temperature, equipment health, and overall processes to take preemptive actions.
Oil and gas industry is another industry which has a significant number of IoT devices and sensors. Oil and gas rigs deploy over 10,000 sensors and all these data are accumulated and transmitted to a cloud network. Some of the data captured are just meant to signify whether the system is functioning properly or not. In such a scenario, an edge computing system could locally compile and analyze the information and only the overall system report along with the important data could be sent across to a cloud network at the end of the day reducing the data traversing the network.
Autonomous vehicle industry would also be an early adopter due to its need for real-time decision making functionalities in self-driving vehicles. With improvements in AI algorithms, sensor technology and computing capabilities, companies like Waymo, Tesla and Audi among others are investing heavily on autonomous vehicles. These vehicles utilize a wide variety of technology and sensors, such as radar, GPS, LIDAR, odometry etc., to analyze and detect the surroundings, and make decisions on its own, keeping safety a priority, while navigating. Signals from these wide variety of sensors are integrated to estimate the position and type of object (i.e. human beings, animals, other cars etc.), the velocity at which the object is moving and the trajectory of the movements. To merge all these sensor information, companies implement a technique called sensor fusion.
What is Sensor Fusion Technology
Sensor fusion is defined as the technique to combine multiple physical sensor data to generate accurate ground truth even though each individual sensor might be unreliable on its own.
Data from multiple sources help remove errors and combining these data with contextual information makes the data more useful than data from a single sensor source. Sensor fusion techniques help remove temporal, noisy input and generate a probabilistically better estimate of the kinetic state of the object. But currently most of the processing of these sensor fusion data is done on cloud and thus suffers from the inherent speed of light latency. An autonomous vehicle has to respond immediately if it sees a pedestrian jumping onto the path of the vehicle. Its takes around 100 milliseconds for a large dataset to travel back and forth from a cloud and a lapse of a fraction of a second might be the difference between colliding with the pedestrian or avoiding the collision. For autonomous vehicles to achieve situational awareness, they need to process sensor information on the edge. While sending and receiving information from the cloud is required, autonomous vehicles need edge computing capabilities for greater acceptance.
Understanding Edge Computing
Edge computing is a distributed open IT architecture that enables systems to compute data near or at the source of information rather than relaying the information to the cloud. Edge computing enables real-time data processing without latency.
With edge computing capabilities, systems can perform efficient data processing as large amount of data can be processed at or near the source thereby reducing internet bandwidth usage. Apart from reducing the overall cost, it ensures the systems can operate in remote locations as well. Additionally, eliminating the need of relaying all the information to public cloud enables an additional security of sensitive information.
The global edge computing market might reach up to $6.72B by 2022 as per CB Insights1. Adoption of edge computing would be based upon:
- Latency minimization between data capture and computation
- Reducing internet bandwidth consumption
- Improving data security by eliminating sensitive data transfer to public cloud infrastructure
- Lowering of operational cost
- Adhering to various regulation and compliance norms with regard to transfer of information across borders
Though edge computing is relatively nascent, major cloud service providers are adding edge computing solutions to their offerings. AWS Greengrass from Amazon enables devices to act locally on the data while still using cloud for management, analytics and durable storage. Azure IoT solution from Microsoft extends cloud analytics to edge devices and can be used offline. Similarly, Cloud IoT Edge from Google extends powerful data processing and machine learning to edge devices.
Infosys Believes Edge Computing Will Bring New Solutions to Old Challenges
Infosys in partnership with Huawei released a smart industrial robots’ solution based on the open edge computing IoT. The solution supports the interconnection of industrial robots from multiple vendors as well as helping manufacturers to anticipate faults and improve maintenance efficiency. The solution is able to reduce the industrial robots’ downtime by over 70% and defect rate by 40%. The solution is also able to schedule production rates based on resource utilization as well as optimize these production lines for maximum efficiency.
Infosys is also working on drone led inspection system having edge computing capabilities. The solution would enable clients to optimally utilize drone fleets to inspect assets in remote location with intermittent internet connectivity without conceding on the effectiveness of the system.
Increased adoption of IoT and reduction in prices have opened up business opportunities not only in the IoT space but also in the edge computing and sensor fusion area. There would be newer business models evolving along the lines of providing sensor fusion platform services besides building computing networks closer to the source of data. Investments in these areas both in terms of money and human resources would enable organizations to stay ahead of the evolution and reap rewards.