Transportation is moving beyond the era of the automobile toward a more fragmented ecosystem of mobility devices, a shift that represents not only a technological upgrade but also a broader transformation in urban life.
Kingwaytek Technology said that as AI matures, cities are likely to adopt on-demand, route-free, schedule-free intelligent shuttle systems — vehicles that arrive when called rather than running on fixed lines or timetables. In the company's view, future AI systems may even develop a form of physical awareness, fundamentally reshaping how people interact with space.
Early versions of this shift are already emerging. In the field of public safety, autonomous inspection vehicles equipped with advanced imaging systems are being tested in cities such as Nanjing, where they can detect road defects, damaged guardrails, and infrastructure wear in real time, transmitting data directly back to traffic management systems.
In consumer and logistics applications, smart luggage and camping carriers capable of carrying heavy loads and autonomously following users through visual tracking have already entered the market.
For more specialized use cases, industry executives point to defense and disaster response. Hyundai has developed a wheel system that can rotate 360 degrees and operate without air-filled tires, allowing vehicles to traverse extreme terrain such as battlefields or earthquake zones without the risk of punctures. Meanwhile, research groups, including those at the University of Tokyo, are training humanoid robots to operate conventional vehicles — including manual transmission cars — suggesting that even non-intelligent vehicles could be rendered autonomous through external AI systems.
On a broader scale, some technologists are beginning to describe the emergence of "driverless cities." Toyota is developing Woven City, where hundreds of residents will live in a controlled environment designed to test future mobility systems and AI-driven urban life. In South Korea, the city of Gwangju has announced plans to deploy roughly 200 autonomous vehicles for urban operations. Advocates argue that such large-scale regulatory experimentation is essential for technological breakthroughs.
At the same time, there is a more critical reading of the landscape in Taiwan. While the island possesses strong engineering capabilities, analysts note that its smaller market size and regulatory constraints may limit large-scale deployment. The effectiveness of AI-driven navigation systems, they argue, depends less on static traffic signals and more on massive, real-time datasets generated by fleets of connected vehicles.
The underlying technology is also evolving rapidly. According to Kingwaytek executives, autonomous driving systems are shifting away from traditional rule-based code toward end-to-end models and so-called "world models." A key emerging concept is "physical AI," in which sensor systems develop an understanding of geometry and physics rather than simply recognizing images.
For example, when a vehicle detects water or ice on the road, it may infer changes in surface friction and predict potential skidding risk, moving beyond visual recognition into what some researchers describe as a form of physical reasoning.
The industry insider says the next stage of development will combine vision-language-action (VLA) models with world models, enabling AI systems not only to process vast computational workloads but also to interpret human intent and emotion through natural language. This convergence, they argue, could bridge the gap between engineering logic and the complexity of human social behavior.
Despite challenges around computing costs and the opacity of AI decision-making systems, many in the industry believe the shift toward new mobility ecosystems is already underway and increasingly difficult to reverse.
Article translated by Elaine Chen and edited by Jack Wu
