Overview

As a Research Assistant in the Avionics Engineering Center at Ohio University, I contributed to the design and evaluation of autonomous navigation and mapping systems for Unmanned Aircraft Systems (UAS) across a range of challenging operational environments — from GNSS-degraded forest canopy to indoor maintenance tunnels and urban canyon multi-copter formations. The work spanned custom hardware integration, algorithm development, flight testing, and collaborative multi-institution research.

Research Thread 1: Under-the-Canopy Forest Navigation

Traditional GNSS positioning becomes unreliable under dense forest canopy due to signal attenuation, multipath, and diffraction from foliage. This project developed a Simultaneous Navigation and Mapping (SNAM) method for multi-copter UAS operating in these environments.

Platform

A 3DR hexacopter was equipped with multiple 2D laser scanners (forward-pointing and downward-slanted), cameras, a GNSS receiver (u-blox M8T), and a low-cost IMU. All data was logged and processed in ROS on a FitPC onboard computer.

Navigation Method

The hierarchical GNC framework integrates:

• Tightly-coupled GNSS/INS (carrier-phase-based Complementary Kalman Filter) when satellites are visible
• ICP (Iterative Closest Point) scan-matching with IMU aiding for local 3D pose estimation from tree-trunk features
• Dual-scanner altitude estimation — the downward-slanted scanner generates a terrain reference, and the downward-pointing scanner matches against it to estimate height above ground
• Barometric altimeter for height calibration in GNSS-available segments

Flight Tests

Two test flights were conducted in a temperate forest near Athens, Ohio (April and June 2015). The more challenging June flight (denser foliage) demonstrated:

• Successful 3D position and attitude estimation consistent with low-accuracy GPS tracks
• Generation of 2D tree feature maps
• Observed multipath effects at forest entry/exit, with reduced noise in open-sky segments

[PDF] A Navigation and Mapping Method for UAS during Under-the-Canopy Forest Operations — IEEE/AIAA DASC 2016

Research Thread 2: Multi-Copter Relative Navigation and Collision Avoidance

In collaboration with a multi-institution team, this work developed a real-time relative navigation capability for multi-copter platforms to enable collision avoidance in urban environments where cooperative vehicles share airspace.

System Design

• Two multi-copters and one ground vehicle, each equipped with GNSS, IMU, and a communication link
• Relative navigation algorithms computed each vehicle's position and velocity relative to neighbors
• Collision avoidance logic implemented on embedded ROS processors — reactive, low-latency
• Flight tests evaluated multiple navigation and avoidance scenarios including close-proximity approach and departure

[Paper] A Real-time Relative Navigation Capability for Multi-copter Collision Avoidance — ION ITM 2016

Research Thread 3: Custom Hardware and Alternative Positioning

A recurring theme across projects was the need for reliable positioning when GNSS is unavailable. Contributions in this area included:

• Custom sensor integration boards — designed hardware to interface LiDAR, depth cameras, and thermal sensors with lightweight onboard computers (Raspberry Pi, Odroid)
• NextNav urban positioning — collaborated with NextNav to integrate their proprietary indoor/urban positioning solution on multi-rotor UAS, providing GPS-independent waypoint navigation in urban canyon environments
• SVD-based plane estimation — implemented a Singular Value Decomposition algorithm for rapid planar surface extraction from point clouds, serving as a key primitive for both localization and mapping algorithms
• UAV dataset collection — designed and executed flight campaigns collecting GNSS, IMU, LiDAR, and camera data in GNSS-denied and GNSS-challenged conditions (urban canyons, indoor, under canopy)