Recently, Nature published an article to introduce the progress of humans like artificial general intelligence (AGI), which was realized by developing a hybrid chip (Tianjic) that supports both computer-based machine learning algorithm and the neuroscience-oriented schemes to simulate a brain. By combining the building blocks for both artificial neural networks and biologic networks, the chip can integrate multimodal information, run through a variety of networks and reach prompt decisions.
As an experiment, The Tianjic research team demonstrated a self-driving bicycle that can successfully and autonomously move around using voice commands. The bike has real-time detection of objects, can follow the human at an appropriate speed, and ride over speed bumps. The diagram at the bottom illustrates the construction of the motors, Tianjic chi, and various sensors such as IMU, camera, audio and etc. In an experiment, the Tianjic-powered bike smoothly performed all the tasks. This is a huge leap toward AGI development.
The real achievement is not the self-driving bike itself, but the chip that is being used to control the bike. The chip was completely designed by the university students themselves. The design process is extremely complicated since chips usually have billions of transistors and other various control units that must be able to communicate at extreme speeds to accomplish tasks. The chip also uses a special type of neural network that models the functions of the human brain, allowing it to maintain balance, control the bike, and also recognize and avoid obstacles.
Uber also announced earlier this year that it is developing a self-driving prototype of a bicycle and/or scooter. This bike has a very bright future if business and technology could evolve in harmony together!
- Nature, Volume 572, Issue 7767, 1 August 2019