Research in MAVI

There are two major research areas in which MAVI is being used as a prototype platform.

  • Design Space Exploration
  • When it comes to building complex systems such as MAVI, it is not known in advance which hardware configurations would be the most suited for the set of applications being targeted. In fact, the designer usually encounters a huge design space from which one has to choose the components and their operating points for the system. This decision, typically, is based on some approximate spreadsheet calculations or just the designer's/team's experience. We aim to create a framework to enable application metrics based exploration of the design space.

  • Sensor Fusion for System Optimization
  • Complex systems have a great variety of sensors for capturing the analog world. However, most of the sensors perform well only in a limited set of conditions. Therefore, it becomes necessary for the system to be able to choose which sensors should be used in what scenario. For example, RGB Cameras provide low quality imaging in dark conditions while Depth images formed through IR scans are very defective in bright sunlight. If we were to keep all the sensors working at the same time, it would be a terrible waste of energy. Therefore, use of context and real time conditions can be useful in selection of sensors which can be used in the specific environment.


    1. Kedia, Rajesh*, Anupam Sobti*, Mukund Rungta, Sarvesh Chandoliya, Akhil Soni, Anil Kumar Meena, Chrystle Myrna Lobo, Richa Verma, M. Balakrishnan, and Chetan Arora. "MAVI: Mobility Assistant for Visually Impaired with Optional Use of Local and Cloud Resources." In 2019 32nd International Conference on VLSI Design and 2019 18th International Conference on Embedded Systems (VLSID), pp. 227-232. IEEE, 2019.

    *Equal Contribution

    2. Sobti, Anupam, Chetan Arora, and M. Balakrishnan. "Object Detection in Real-Time Systems: Going Beyond Precision." In 2018 IEEE Winter Conference on Applications of Computer Vision (WACV), pp. 1020-1028. IEEE, 2018.

    3. Kedia, Rajesh, K. K. Yoosuf, Pappireddy Dedeepya, Munib Fazal, Chetan Arora, and M. Balakrishnan. "Mavi: An embedded device to assist mobility of visually impaired." In 2017 30th International Conference on VLSI Design and 2017 16th International Conference on Embedded Systems (VLSID), pp. 213-218. IEEE, 2017.