An Intelligent System to Sense Textual Cues for Location Assistance in Autonomous Vehicles

Area of Science:

• Computer Science
• Artificial Intelligence
• Robotics

Background:

• The rapid advancement of autonomous vehicle technology necessitates enhanced driver assistance systems (ADAS).
• Current systems require improved methods for environmental perception and navigation assistance.
• Bridging the communication gap in autonomous vehicles is crucial for safety and reliability.

Purpose of the Study:

• To explore the utility of environmental textual cues for location identification in autonomous vehicles.
• To develop a system that assists drivers in finding desired locations and stopping points.
• To enhance the capabilities of advanced driver assistance systems (ADAS) towards level-5 autonomy.

Main Methods:

• Utilizing natural language processing (NLP) to sense textual cues from the vehicle's surroundings.
• Implementing similarity learning to match driver-input keywords with environmental text.
• Developing a system that detects, informs, and controls vehicle speed for stopping at relevant locations.

Main Results:

• Demonstrated the system's efficiency and accuracy in identifying desired locations using textual cues.
• Validated the proposed approach on four benchmark datasets.
• Showcased the system's ability to assist drivers in autonomous navigation and stopping maneuvers.

Conclusions:

• Environmental textual cues can be effectively leveraged for location-based assistance in autonomous vehicles.
• The proposed NLP and similarity learning approach enhances the functionality of ADAS.
• This research contributes to the development of more reliable and intuitive autonomous driving experiences.