Shrutarv Awasthi

Goal: Enable high-speed robots to navigate in dynamic industrial environments. • Designed a multi-camera system comprising two event cameras and RGB camera and callibrated it. • Created a multi-modal, multi-task dataset for 3D object detection and motion segmentation. • Bench-marked a Recurrent Vision Transformer on the dataset for moving object detection, achieving 75% AP. • Collaborated with industry partners to align research with real-world requirements.

Project: Drone Trust (Indoor Logistics) A comprehensive framework and digital-twin simulation is developed to demonstrate the feasibility of UAV swarms performing logistics tasks such as transport, guiding humans and inspection in constrained warehouse settings. In parallel, machine learning methods for human activity recognition are explored. A dataset consisting of human UAV interaction in a warehouse setting is also published on Zenodo. [ICPR 2022, Logistics Reasearch Journal, IFAC 2025, MHCL, AiSOLA]

Preparing and implementing unit test cases for context dependent learning using deep reinforcement learning in python.

- Generated multivariate time series data using conditional GANs. - Time Series Classification using deep learning architectures. - Implemented MetaBags: Bagged decision trees for regression

Forecasting weather pattern in Germany by performing: o Data aggregation and data mining. o Data exploration: Identify relations between features using python plots. o Data Analysis: Visualizing data to identify trends, creating & selecting features. o Creating a model which fits the data and makes highly accurate prediction. o Automatic data extraction using python o Planning a strategy considering multiple factors for electric vehicle charging stations along German motorways.

- Developed a feedback functionality for the company’s website using JAVA (GWT) - Implemented statistical analysis of user inputs. - Created style sheets (CSS) for designing widgets. - Designed HTML widgets.

• Coordinated with client in gap analysis & solution mapping of Product Lifecycle Management (PLM) application-Teamcenter, based on JAVA. Helped clients in mapping business requirements by customizing multiple functionalities, in the vanilla version of application. • Data analysis and software development. • Automated using LINUX scripting. • Used SQL to query the database while bug fixing & customization. • Used python to analyse and forecast application usage by users. • Ensured quality of deliverables.

Computer Vision, Deep Learning, Artificial neural networks, computational intelligence, mobile robots, robotic manipulators, Mathematics

Industrial automation, soft computing techniques, micro-controllers, control system