[Click for Resume] I am an ECE Master's student at UCSD studying signal processing. I am concurrently employed at Synaptics inc. as a BT systems development intern doing firmware/RTOS development as well as working on BLE's Channel Sounding functionality. I have done research at Prof. Dinesh Bharadia's lab, WCSNG, on the localization team. I focused on developing novel sensing platforms to improve state estimation techniques in robotics. In particular, I focus on using RF-sensing modalities for increased robustness and information aquisition in the SLAM (Simultaneous Localization and Mapping) problem. I am currently looking for an internship for Summer 2024 in the field of wireless communications and sensing.




Our goal is to mainstream the use of Wi-Fi sensing in robotics. Wi-Fi provides numerous advantages as a sensor, chiefly that Wi-Fi signals come pre-labeled with a MAC address, meaning that, unlike a visual sensor, no cross-correlation or loop closure detection needs to be performed. My work focuses on developing Wi-Fi sensor frameworks and integrating them into SLAM (Simultaneous Localization and Mapping) systems.


  • P2SLAM: Bearing-based WiFi SLAM for Indoor Robots (RA-L 2022)

    Aditya Arun, Roshan Ayyalasomayajula, William Hunter, and Dinesh Bharadia

    While previous Wi-Fi SLAM systems have used simple measurements like signal strength, we show that bearing estimates computed from incoming Wi-Fi signal's CSI (Channel State Information) can vastly improve a Wi-Fi sensor's performance. We demonstrate that a-posteriori trajectory estimates on par with a state-of-the-art SLAM system like RTAB-Map can be retrieved using just wheel odometry and CSI information at the robot, given realistic indoor Wi-Fi infrastructure deployment.


  • ViWiD: Leveraging WiFi for Robust and Resource-Efficient SLAM

    Aditya Arun, William Hunter, Roshan Ayyalasomayajula, and Dinesh Bharadia

    Visual SLAM systems need to keep a dictionary of already-visited locations so that they can detect when they have re-visited areas in order to correct their map. We show that incorporating Wi-Fi channel information into the SLAM system can effectively remove this need in places where Wi-Fi infrastructure is deployed. This is because Wi-Fi transmitters in the environment can be treated as global landmarks that keep the robot's pose anchored to a consistent frame. We develop a lightweight wrapper that can replace Kimera's global pose graph optimizer, reducing compute and memory usage while attaining similar performance.


  • RF Tools for Robotics

    I am currently developing ROS (Robot Operating System) tools to enable Wi-Fi and UWB sensing platforms for robotics applications.


  • wiros_csi_node

    Integrates the Nexmon CSI platform with ROS in order to extract and collate wireless channel data.

  • wiros_processing_node

    Implements state-of-the-art AoA algorithms to create camera-like sensing capabilties from extracted channel data.

  • A map of strongest AP connectivity across a robot's trajectory (colored lines) and estimated AP locations (Diamonds) by WiROS over a LiDAR map generated by Google Cartographer.