Deep anticipation: Lightweight intelligent mobile sensing for unmanned vehicles in IoT by recurrent architecture

Guang Chen, Shu Liu, Kejia Ren, Zhongnan Qu, Changhong Fu, Gereon Hinz, Alois Knoll

Research output: Contribution to journalArticlepeer-review

2 Scopus citations


Integrating Internet of things (IoT) techniques into automated vehicles has been a vision in intelligent transportation system, there is however seldom researches addressing it. To this end, we envision a scenario: Short-range on-board sensor perception system attached to individual mobile applications such as vehicles are connected via IoT and transferred to longrange mobile-sensing perception system, which can be used as part of a more extensive intelligent system surveilling the environment. However, the mobile sensing perception brings new challenges for how to efficiently analyse and intelligently interpret the deluge of IoT data in mission-critical services. Among these challenges, one bottelneck is the quality of service of IoT communication. In this article, we model the communication challenge as latency, packet delay variation and measurement noise which severely deteriorate the reliability and quality of IoT data. We propose a novel architecture that leverages recurrent neural networks and Kalman filtering to anticipate motions and interactions between objects. The model learns to develop a biased belief between prediction and measurement in different situations. We validate our neural architecture with synthetic and real-world datasets with noise that mimics the challenges of IoT communications. The proposed neural architecture outperforms state-of-the-art work in both computation time and model complexity.

Original languageEnglish
Pages (from-to)1468-1474
Number of pages7
JournalIET Intelligent Transport Systems
Issue number10
StatePublished - 1 Oct 2019


Dive into the research topics of 'Deep anticipation: Lightweight intelligent mobile sensing for unmanned vehicles in IoT by recurrent architecture'. Together they form a unique fingerprint.

Cite this