# Memory-Based Autonomous Driving KEPT Uses Past Driving Scenarios to Improve Real-Time Decisions By Mursal Rahman | Written with AI assistance | Published 2026-04-22 | Tech Source: Trend Hunter, https://www.trendhunter.com/trends/kept References: [digitaltrends](https://www.digitaltrends.com/cool-tech/this-ai-lets-self-driving-cars-remember-past-drives-to-plan-safer-routes/) & [eurekalert.org](https://www.eurekalert.org/news-releases/1124196)  The KEPT system, developed by researchers at Tongji University and collaborators, introduces a new approach to [autonomous driving](https://www.trendhunter.com/trends/nuro-r2) by enabling vehicles to reference past driving scenarios when making real-time decisions. Instead of relying solely on immediate visual input, the system retrieves similar road situations from a database and uses them to guide safer and more accurate [trajectory predictions](https://www.trendhunter.com/trends/Experience-trained-AI-systems). This allows self-driving systems to better navigate complex and unpredictable urban environments. For businesses, this approach could improve the reliability and safety of autonomous vehicles, addressing a key barrier to adoption. Automotive companies and mobility platforms may benefit from reduced accident risks, increased consumer trust, and more scalable deployment in dense city settings. It also signals a shift toward more context-aware AI systems, creating opportunities for similar applications in [robotics, logistics, and smart transportation](https://www.trendhunter.com/trends/Autonomous-supply-chains) where decision-making relies on both real-time data and learned experience. Image Credit: Iv-olga / Shutterstock.com ## Trend Insights (Trend Hunter) - Score: 4.6/10 - Popularity: 32% | Activity: 35% | Freshness: 70% - Audience gender: 50% men, 50% women - Primary generations: Millennial, Gen X - Top markets: Asia ## Categories [Trend Hunter](https://www.trendhunter.com/trends) > [Tech](https://www.trendhunter.com/tech) > [Science](https://www.trendhunter.com/science) ## Key Themes ### Key Themes Behind This Trend - **Memory-augmented Perception:** A system that augments real-time sensor input with archived driving scenarios to improve recognition of occluded objects and rare events. - **Context-aware Trajectory Planning:** An approach where trajectory predictions incorporate similar past maneuvers to produce smoother, more confident navigation in complex urban layouts. - **Experience-driven Safety Validation:** A paradigm that leverages historical incident and near-miss data to benchmark and validate autonomous decision-making under diverse edge cases. ### Where This Applies - **Automotive-manufacturing:** Vehicles embedded with memory-retrieval modules could exhibit higher baseline safety performance, shifting OEM value toward software-rich platforms. - **Urban-mobility-platforms:** Ride-hailing and shared-mobility services could use scenario-based decisioning to increase reliability in dense city operations and improve rider trust. - **Logistics-and-warehouse-robotics:** Autonomous material-handling systems informed by past navigation episodes may better handle dynamic human interactions and congested aisles. ## Related on Trend Hunter - [University-Focused AI Governing Frameworks](https://www.trendhunter.com/trends/tsinghua-university.md) - [Autonomous Vehicle Platforms](https://www.trendhunter.com/trends/tensor-robocar.md) - [Friendly Efficient Delivery Robots](https://www.trendhunter.com/trends/pat-autonomous-delivery-robot.md) - [AI-Powered Driving Apps](https://www.trendhunter.com/trends/aipowered-driving-app.md) - [Safety-Focused Automotive Technologies](https://www.trendhunter.com/trends/auto-shanghai-2025.md) - [Groundbreaking Automated Driving Systems](https://www.trendhunter.com/trends/automated-driving-system.md) - [Self-Driving AI-Backed Models](https://www.trendhunter.com/trends/alpamayo-r1.md) - [Cognitive Workflow Engines](https://www.trendhunter.com/trends/layer-brain.md) - [Intuition-Inspired Robot Safety Technologies](https://www.trendhunter.com/trends/predictive-safety-engine.md)