Hello everyone. My name is Ravi Kala, and first of all thank you for giving an opportunity to introduce myself and then conference opportunity. So if you have any questions regarding this topic, you can reach out to me over the email or and link it in. So today I'm going to discuss in this conferences the evaluation of. A enhanced automotive infotainment system and that and as well as the, I'm going to discuss the technical innovations and then future prospects. So let's start with the, some basic things. So you are, automotive entertainment? Automotive entertainment is information plus entertainment is entertainment. When you are, when you sit in a car, you see the display in front of your old driving suite or in front of, in the console. That is the infotainment system. You can listen the songs and then, you can entertain the, things like you could do the video games or in a parking, in a park mode. And then as well as the, you can do vehicle a operations, like change settings or that is the info system. So hope you are, you know that. But yeah just wanted to give you brief overview, but in this topic I'm going to discuss how the AI. Is influential, the info interacting with the infotainment system, automotive infotainment system. So in this conference, I'm going to discuss the technical framework and challenges and as well as the breakthrough solutions driving the development of AA power automotive infotainment systems. So let's start with some basic thing like the market growth and technological evaluations. So if you see this there are the 2023. Around 21 part six, 3 billion revenue from infotainment a infotainment so that, that use that will be projected like 37 part 6 billion by 2023. A enhance system, which is currently accounted for like around 42.3 of the market. That is expected to dominate like, more dominate the future, grow future automotive industry. For the a. But the basic thing is bias recognition because of, you need to talk with you need to talk to the infotainment system or that is the easiest and more, more fastest way to communicate with your infotainment system. So in 2020 2020, the the wise accuracy, it used to be like an 82.4 percentage around around 20 wise commands. Almost three or two operations. Operations, like in a wise commands got failed able to process that command. But in 2023 that increase to 96.2, right now it's almost 99.99 percentage accuracy of the, of your wise commands. And then the, and the other thing is the fast response, the responses also very reduced a right now it's 2.7 seconds to get. Get your answers whatever you are asking to the entertainment system it'll respond like that. And then and then the there are so many things involved we will get into that things first of all computational architecture evolution. So in 2020 in 2080, let's start with the 20 28, 20 18. So we used to like in automotive industries, used 1, 2, 2 processing means we call the vehicle interface process. The vehicle is. Consisting of ecs, electronic control units, each con, each ECU consisting of specific set of operations BC body control module. It's specifically designed for braking system or lighting or some of the most critical operation. And then there are lightning module, especially for lights, lightning system. And then there are like, vehicle vehicle control units and then as well as the like you audio amplifier amplifier modules. Amplifier modules are like, audio systems. Those kind of things are like, it's consisting of the auto, the vehicle is consisting of number of, so in 20 20 18, it used to be one to two crossing limits, but 2020 we. It increase it to two to four processing units more demand, more more capabilities are introduced. And then 2023 and 2022 at used it was like, five to eight processing units. So imagine 5, 2, 8 is 1 trillion operations per second. The whatever the, the five to eight crossing units. It'll it'll operate 1 trillion operations. And right now we are almost triple that 3 trillion operations per second. That's how we increased the capabilities of the operations and then low power consumption. That is also another another thing we need to we need to consider. All, automotive is first priority is safety. Safety is when you are driving your car safety is the first priority for the automotive industry. That's why. Every automotive industry when they are implementing any. Any any capabilities or anything, the safety is the first reality for their for their operations. So that's the that's how they operating. So let's get into that. Drive driver monitoring and personalization things. Let's the AI is also uses not only for your voice recognization, but also camera. So how the, how you are whether you are sad, whether you are sick, whether you are, based on your eye movements, based on your facial expressions it'll recognize and then say, based on your command if you are a dull if you are, if you got into a trouble or anything, it automatically detect and directly connected to your when you are in got an accident any. Need anything any, any sudden movements, it automatically directly connected to the back office systems, and directly connected to your 9 1 1 to get get help as possible and then car your incidents. That's how this AI is added those capabilities and as well as the. I just I just your information like, your personal information whether it's saved saved, locally, saved in the back office, but it's but it's needs to be private. You. No, no need to be shared with someone. So that is our sort priority for the ai. And then there is an name. Neural networks. That is like analyzing multiple input. You are driving same location multiple times, and then a, recognize that one and then determine okay. You are going to this one you are going to this road, and then is there any traffic in that location? And then it'll give an in input, like no suggestions. Oh, hey you are going into this direction. We have some trouble. We are closing some roads or something like that. So then he, then you know you have any better better things and better schedule in your day. So that's how the, your your time is saved. And then you can, you can you can like in safety of your when you were driving safety is also considered. And then as, as well as the driver attention when you are driving telling me, traffic. We had you as like something. Then it'll recognize that. Then think, okay, hey, there is a traffic. You need to be alert. Then you can do that. That's how, yeah. These things are working. So let's predicting a and context context awareness. There are four things we need to be considered temporal analysis and then geo pattern intelligence and then behavioral pattern, the communication and feedback loop integration. These are designed for a different different, things like, day after week pattern recognization when you are when you are particular day, you, you are doing something like, you are able to do something, then a, predict that one then give the suggestions. And then there is another thing like, you were able to you are a AV. Electrical vehicle, your charging station, recommended charging stations. It'll give the recommended charging station. And then audio setting based on your mood, you can the a give, Hey, I'm going to set up this songs for you. Whether you agree or disagree, it's up to you. But yeah it'll give an suggestion and then behavior pattern based on your mood, based on your your preference based on your media preference. It'll it'll give an a, like an, okay, I'm playing this song and looking, okay I want to play this song, yes or no? Yes. Yeah, you could it will play then without your without, much thinking of your it automatically recognize your mood and then it'll, and then feedback. Look, that is another thing. When you are when you are like, when the a given in suggestion, but you don't like it, it's okay, but. It'll why it is rejected. It'll analyze the ai and then next time, don't repeat that. Mistakes. So that's how it yeah. This AI is working so. And then resource constraints and optimization. In the automotive industry so there are hardware limitations and then power efficiency requirements and body management and safety critical ation. So these are the main constraints for aa advancements right now there are, hardware limitations that like, one of the core hardware in the automotive industry it is to it used to be $1,200, but now it's almost three to $5,000 per need. So that's how and then the budget is also very important for the, and the space restrictions. Also, you need to be small. The electronic control unit, a CU is must be small, and as well as the doing more operation, trillions and trillions of power operation. And then second thing is battery efficiency. That in automotive industry EV ranges question for hey my, my company's doing six 50 miles per full charge. A my, my ev is like 300 miles per full charge within 15 minutes, something like that. No, power efficiency is after thing we need to consider. And then memory management then. So how many operations we are performing to the users, and then we need to show. We need to store and then deploy and then use that, those information, and then give it to the right feedback. Write write information to the try user. And then there is safety critical operation. That is the that is one of the performance determin performance guarantees. Oh, hey, my automotive industry is doing good with this operations. Something like that. So that's how it's it's working. And then there are privacy and data management. This is another thing. So privacy is the main important for for our automotive industry. Whether when you are introducing the a into the automotive industry, because of you share your personal data, because of, you shared your personal data, you shared your things to the, your car and. That needs to be stored somewhere personal. That's how and then there are different things we need to consider as an automotive automotive engineer. F first thing is. And device process only since two data are stored in locally, like in your vehicle, and then the remaining things are like, stored in cloud services and then reduce the privacy exposure to the remaining vendors. And then and the differential privacy. And then noise. Transmission and math mathematically guaranteed individual privacy while maintaining the static statistical usefulness document. These are the things, and then the federated learning that dis distributes model training across the vehicle. Keep the raw data in local, allowing the fleet wide AI operations without centralizing the personal data. So time limited storage hey I'm going to store this data like one month, three months, something like that. When you change the, it'll let automatically delete delete your data, delete your personal data, and then keep the vehicle three without any personal data. That's how privacy. We are going to introduce to the automatically systems. And then safety. As I mentioned before, automotive automotive's automotive is first thing is the safety and distraction. Distraction mitigation. So safety is the first priority. We never, we are implementing anything. Safety is the first degree reality. So what are the medications we are going to? First thing is workload. Adaptive interfaces. When the when the UI elements based, standard driving complexity, reducing the car, congregating calculating load by 43% during hiding, driving situations. And then multimodal interaction design. Whether you are using voice, a touch or architecture anything it need to be no, no need to be distract you and driving. That is another thing. There are there are couple of things we need to consider. Driver must be consultation on the road without without doing any further aggressions. Hey, touching something on the entertainment system. No. We are not doing that one 'cause of, safety is the first the first ity, and then attention and avail interruption. You, if you are talking to Thea, that's okay if you are like, if you are not concentrating and concentrating not concentrating on your driving things then that's the interruption comes into the picture and then we can notify the driver. Hey. We need to be ice on your road, ice on the road. So that's how different things mitigation we are going to implement into that safety structure, destruction, mitigation things. And then the further thing is testing and validation as, how the testing is conducted. So first thing is we are doing some basic basic testing the National Highway Traffic Safety at the administration whatever the guidelines they. They're provided. We do that some basic things and as well as the, some of the more generic principles principle verifications, and then brand specific interaction standards. And then I. Then there is a controlled environment test testing. We are like, we are tested like, only certain things and then certain behavioral analysis and then certain task we need to give to the entertainment system and then perform do observe the behavior. And then further we are we are going to develop that driving simulator validation that is just in in just in, in, in house driving. But you are giving all the feedback to the AA and then how the reaction time, and then how the it how it is working. And then the third, the fourth phase is realtime real world validation. We have in a certain drivers, we will give the vehicles to the owners. We call them test fleet owners, test fleet vehicles. And then the owners will give an a complex instruction to the AA and then how they interacted with that things. And then we can modify, and then we can rectify those things, and then we can improve our aa into the info system. And then future directions and technical group. So in the future, there are things need to be considered in our ai. So first thing is multi modern, deep deep learning. Integration of vision, audio and sensor data and unified AI model AI models, and then autonomous integration so when you are driving with hands free, hands, self driving. So you are, you don't need to be looking into the road or some like fully autonomous. You don't need to be, but semi-autonomous, yes, you need to be sometimes you don't need to be hands on the wheel like autonomous, but in that state, we need to be interacted with that aa to that autonomous inter interaction. So we have a better autonomous system. And then AI documentary reality integration. So that is that is another thing like, wind in the windshield. We can show hey. This is the information, this is the this is the weather, this is the speed you are, this is the, this is the traffic you are going to, in the, in 15 minutes or 10 minutes, something like that. And then emotionally based on your mood, based on your emotional state driver. The AA will give an advice and responded according to your modes. And then V two X vehicle to everything, communication, V two vehicle to network, vehicle to other vehicle to, a charging area, vehicle to interface. So this kind of communication is vehicle to another vehicle. Vehicle to vehicle is also another communication. So how that things are we need to looking into that one. So this is how a we we will use into that automotive industry and then we will focus more on the deep learning models and as well as the, how the. A is interacted with infotainment system and then as well as the autonomous integration into data infotainment system will achieve a better and bright automotive systems. Thank you so much and thank you.