Aditya Pulipaka

Developing ROS-based system that uses Spinnaker SDK to acquire images from ethernet-connected FLIR Blackfly S cameras, performs image-based pose tracking, fuses data from multiple cameras, and outputs global position estimates for people and robots in the GDC atrium.

Leading control algorithm development, developing reinforcement learning pipeline for HEMS system.

Prototyped single and dual-yoke test-rigs for gradual building of PID control algorithm. Used Arduino + custom power delivery system to automate coil winding, saving ~5hr labor per prototype Researched and tested gap height sensing and feedback strategies to optimize levitation stability Worked with MATLAB + Simulink to simulate feedback control of 4-yoke test chassis

Developed end-to-end software resolutions, updating low-level requirements, FCS code, and low-level tests (LLT) to resolve 5+ software problem reports. Automated 3 LLT batches for one test set, eliminating tedious debugging. Suggested high-level test and high-level requirement changes for these problem reports. Drafted redlines low-level requirement documents for software changes being considered for future type certification builds. Made 10+ updates to field maintenance windows application, spanning UI/UX and application backend code. Gained comfort with .NET C# code using WPF. Led development of environment transition tool written in Python to transfer 3 integrated communications databases (ICDBs), 6 communication functions classes, and ~1000 high-level tests from VALID test environment to a proprietary C# environment, saving 6-12 months of manual work.

Developed LiDAR SLAM + map-based localization pipeline, building off Open3D, KISS-SLAM & KISS-ICP tools. Performed C++, Python, PyBind modifications to these tools to fit our needs. Combined ICP and feature-matching to develop localization nodes achieving sub-5cm accuracy. See Lidar SLAM package project under projects section

Developed monocular localization pipeline for Sentry robot using openCV’s SolvePNP tool Tested various odometry methods to achieve greater positioning and movement precision

Stuffed prototype boards, drafted manufacturing work instructions, and drafted minor design changes like charging port placement for the KneeStim rehab device

The BMAL1/CLOCK heterodimer forms a negative feedback loop with the Period/Cryptochrome heterodimer. BMAL1 and CLOCK transcribe Period and Cryptochrome when phosphorylated. Period and Cryptochrome then block phosphorylation of BMAL1/CLOCK, completing the loop. We first purified BMAL1 antibody from immunized egg yolk, then used it to investigate phosphorylation patterns in WT, CRY1KO (Cryptochrome 1 Knockout), CRY2KO, CRY1/2 DKO mice to determine effects of each variant of cryptochrome on circadian transcription-translation feedback loops. We found that both cryptochrome 1 and 2 are capable of maintaining the feedback loop alone, as they match the BMAL1 phosphorylation pattern of Wild Type mice. However, as expected, when both cryptochromes are knocked out, there is a loss of rhythmicity, as BMAL1 remains phosphorylated at all times.

Co-captained our school's robotics team to our first Texas State championship debut. Drafted many components, ranging from battery holders to claw mounts, and perfected many more, from control hub fasteners to game element claws. Helped adapt RoadRunner pathfinding algorithm for our robot, working around size and structural constraints that were prohibitive to additional sensing. Participated in community outreach activities including Greenhill's Teaching CS club, where I aided with the help desk and uploaded CS tutorials to our YouTube channel.