High capacity batteries key to evolving autonomous systems
Cloud computing supports a vast array of information and systems every day. Whether it’s Google, social media or file storage, billions of people rely on the internet to stay connected. When that connection is interrupted, it causes inconvenience for the average user. However, if autonomous systems were to rely solely on cloud computing, maintaining a connection could mean the difference between safe operation and disaster. Imagine a self-driving car that required a stable cellular connection to avoid collision.
For federal agencies, cloud computing presents additional security concerns surrounding the possibility of data being intercepted in transmission. Heavy encryption requires high computational power, which in turn requires a great deal of energy to operate. Communicating with the cloud can also cause delays in response time, as information is passed from the site to the computational data center and back again. Mission-critical operations rely on edge computing to process time-sensitive data. The trade-off is that high volume edge computing — computation that occurs at the local device — requires strong power supplies, often in the form of thousands of batteries.
“Battery technology in the United States is way behind in terms of manufacturing,” said Vikas Tomar, professor of aeronautics and astronautics. “The problem is that current battery charge capacity is very low, in order to maintain their stability and make them safer. But autonomous devices require high capacity batteries to operate.”
The optimal combination for autonomous devices is a hybrid of cloud and edge computing. At present, high voltage batteries needed for efficient edge computing are costly and unstable. In collaboration with AAE Associate Professor Dengfeng Sun, Tomar uses machine learning to measure the efficacy and stability of high capacity batteries using sensors the team has developed.
The team incorporates management systems to collect data on the performance of high capacity batteries to determine how they fail and how that failure can be managed. The systems will also review the manufacturer’s history and ideal operational conditions of the device to make a determination of the safety of the system and how much it should be charging.
“The balance of edge and cloud computing is very important and varies among systems,” Tomar said. “Do you need quick decisions? How many systems are there? Are you dealing with thousands of devices vs. one device? The key thing is to improve energy device security based on the machine learning that we have developed.”
In addition to extreme conditions, devices in orbit, such as satellites, are also threated by collisions with space debris. Most of these devices are equipped with heavy shields to protect them, which add significant weight to the object, requiring more power to support it.
“The longer autonomous devices operate in extreme conditions, the more energy they require to perform,” Tomar said. “The current voltage capacity for batteries in space is very low and they require solar energy to recharge. Our approach, based on a combination of cloud and edge computing, will enable devices, such as satellites, to operate in darker places for longer and to store more charge so they can perform more operations.”
This story appeared in the Fall 2020 issue of Aerogram magazine.
