Digital Threshold Live Episode 4 Blog – Why Technology Convergence in the Digital Threshold Matters
The continued acceleration of the digital transformation has unleashed seemingly limitless possibilities for technological applications, from the widespread global standpoint all the way down to the personal level. The emergence of artificial intelligence (AI) presents incredible opportunities in many arenas, including the practical application of physical security.
Anil Chitkara, Evolv Technology Co-founder and host of Digital Threshold Live, was joined by Mahesh Saptharishi during episode No. 4 to discuss the technological possibilities at the intersection of sensors and AI. Saptharishi is the CTO of Motorola Solutions and leads innovation of mission-critical communications, as well as video and command center software.
Saptharishi provided a detailed and thorough perspective on the synergies of machine learning, AI, big data and analytics and how each plays a necessary part in the digital threshold and the creation of state-of-the-art physical security systems.
“Machine learning are the core algorithmic capabilities that power AI,” Saptharishi said. With regards to physical security, “when cameras, or when systems, see things, detect objects or respond to what the objects are doing in the scene, that is artificial intelligence, but that ability to detect and the ability for that system to adapt to the environment is powered by machine learning algorithms.”
Episode 4 Highlights and Provides Insights on the Security Threshold
Saptharishi explained that AI has come a long way in the past 10 years as computing power and speed have taken major strides – in large part due to the evolution of gaming GPUs.
“I think storage becoming cheaper, network bandwidth becoming cheaper, the ability to collect data becoming more practical – that acted as a fuel that powered all these algorithms to develop further and actually reach their performance potential and become practical through the processor technology that has come out,“ Saptharishi said.
This has been instrumental in AI development. One such real world application is AI understanding audio and speech patterns, as well as analyzing video. Utilizing improved GPU technology, these computations can now be done in real time. What once was done with multiple computers can now be done from your phone.
Chitkara asked about the human impacts of AI and whether this technology is replacing people or helping them.
To a degree, AI can replace or augment existing jobs done by humans.
“But, that said, humans are not static entities in terms of how we apply our intelligence,” Saptharishi said. When AI replaces certain tasks, people can focus on other areas where AI does not apply presently or perhaps cannot perform in the same way a human would.
For example, AI can assist law enforcement by helping search for people on video or by calling the attention of the officer to a particular situation that would require human judgement on whether or not intervention is applicable.
Saptharishi also explained the process of developing AI for particular applications. It starts by identifying the human factor opportunities. By shadowing individuals as they progress through their normal tasks, the development team can determine tasks that can be automated or assisted through technology.
This can greatly increase the productivity of the individual. With the automation of sensory actions, people can do things earlier, faster and also allow for more response time to particular situations, as technology has assisted in collecting information in an expedited manner.
In terms of security and technology, Saptharishi noted some key trends that have emerged. One trend is the combination of sensing modalities to create more powerful solutions. Other trends are the increased utilization of cloud connectivity and the integration of public safety, private security and enterprise security.
Chitkara and Saptharishi also discussed the security threshold and the factors that make AI successful for this application.
“Along with this notion that you need a high throughput solution, the threshold cannot become a bottleneck,” Saptharishi said.
The technology threshold needs to be as seamless as possible and not become an overwhelming burden to the flow of people. Secondly, the threshold needs to identify a person along with the right context. Is this person permitted to enter? Along with that identity permission, is this person bringing along something that is not permitted, such as a weapon or an illness?
“The days of somebody sitting in front of a security operations center, watching a video wall, hoping that they can detect something that is potentially suspicious or requires attention – I think those days are starting to reduce,” Saptharishi said.
Watch the full episode OnDemand now:
Too busy to sit and watch a webcast? Listen to our podcast:
Watch Previous Digital Threshold Live Episodes OnDemand.