Here is the scenario. The driver opens the door and starts the engine, and the Persona nicely greets: "Good morning Vojtech. I hope you are doing well?" Persona promotes an optimistic attitude, and the driver replies: "Fantastic, I am ready to have a great day."
Enjoy Safe Driving
Driving Experience is a voice-first application developed with the Flowstorm platform. We have selected the Car Companion as an example application showing the latest technological advances. The demo introduces an essential component, the Intelligent Persona (IP), accompanying the driver and the passengers. The Persona’s conversational abilities, context, association, and reasoning are critical for Conversational AI applications. The Flowstorm Platform offers all the necessary tools to design and deploy complex Persona applications.
We wanted to design the Persona to become the best driver's companion, but good companions need to know their bodies. Therefore, we have equipped the Persona with a Knowledge Database (KDB) to store the driver's profile and other useful information. The best moment to learn the driver's initial profile is during onboarding. To make the Persona helpful, we also ask the driver to share their calendar, contacts, and other personal details. Furthermore, the Persona remembers the driving history, destinations, driving habits, passengers, etc.
Now the journey begins. After leaving the parking lot, the Persona proactively checks the driver's calendar. It is Sunday, and Vojtech has "Grandma visit" on his schedule. The Persona knows that Vojtech has two grandmas; Daisy and Rosemary, and from contacts, it knows where they live and what their phone numbers are. Thus the Persona starts the next conversation by saying: "I can see in your calendar that you plan to visit your grandma. So are we going to her place now?" As the Persona knows that Vojtech has two grandmas, it needs to disambiguate, asking: "Is it Daisy or Rosemary?" The driver responds and the Persona then confirms that it's grandma Daisy and sets her address: "I'll set Daisy's address to the navigation."
The persona also gives notifications, for example when the car is running out of petrol. In such a case it suggests the nearest gas station. The driver might ask the Persona to remember a reminder: "When we are there, remind me to buy windshield washer fluid." The Persona saves the item and acknowledges: "Sure thing, I will remind you to buy windshield washer fluid."
The next dialogue makes use of advanced personalization. The Persona remembers the previous trip to grandma Daisy and notices that the driver stopped at a florist last time. The Persona, therefore, continues: "When you last visited grandma Daisy, you stopped to buy her a bunch of flowers. There is a new florist shop along our route. Would you like to stop there?" The Persona has access to several web services. It is then easy to find the nearest store and get the rating. When the driver asks: "Is this shop any good?" The Persona quickly replies: "It has a very good rating of 4.5 out of 5 stars."
In the next dialogue, the Persona notifies the driver about slow traffic ahead and suggests texting Daisy the estimated arrival time: "There is a car accident. We will arrive about 50 minutes later." The Persona is connected to the driver's cell phone and conveniently asks: "Should I text your grandmother about it?" The driver dictates the text and, before sending it, the Persona replays it.
The Persona also observes the driving conditions, which may lead to the following message: "This seems like a rather monotonous driving situation. How about some Rock'n'roll music to help you stay sharp?"
A few minutes before the arrival, the driver asks the IP: "Where can I park at my destination?" Before the Persona replies, it checks the payment habit from the profile to announce: "There is only one parking option available. I know you usually pay by credit card, but that is not possible here." Here the Association Manager is activated suggesting using an ATM: "But if you do not have cash, there is an ATM just 20 meters away from the parking spot."
Before approaching the destination, the Persona checks the weather to inform the driver that it may rain: "The radar is telling me that it might start raining soon."
The Persona has access to the car sensors, and therefore, it may recognize the passenger. The persona changes the greeting: "Good morning Vojtech and Naomi," and continues asking about the destination: "Are we going to Vojtech's work like last Monday?" And now, the passenger takes over, replying: "We are actually going to my place." The Persona again repeats the name and location to let the passenger correct it in case of speech recognition problems etc. "All right! Let's go to Naomi's home."
While driving to the next destination, the driver asks about the latest sports news: "And tell me the sports news during the drive, please." The IP starts playing a preselected sports channel. Another Association Manager capability is to suggest better parking options based on the traffic situation: "Vojtech, there's a traffic jam ahead. So it might be faster to park in this neighborhood and take the subway to the center."
The next part of the video shows the driver's request: "What can I do tonight?" The IP again refers to the driver's profile and finds the top two events: "There is a concert of your favorite band XYZ at 8 PM." The second event is: "Since you like soccer, you might also enjoy watching the Champions League game between AS Barcelona and AC Milan at 9 PM." In reply, the driver asks the IP to remind him before the game starts. And the IP assures the driver by saying: "I will send a message before it starts." Consequently, the driver will receive a reminder on his cell phone.
The demo shows several practical situations we can experience every day. The Persona has to be adaptive and needs to remember personal information. It also collects the history and context. The Persona also has access to Internet databases with dynamically changing data. The essential part of the Persona is a dialogue management system controlling all decision-making and integrating other AI features.