28 July 2022
—— 8:30 AM to 9 AM – Collecting registration kits and spot registrations ——
8:45 AM – 9AM
9 AM – 10 AM
The only real-time measure we have for investigating fetal wellbeing is the fetal heart rate, which forms specific patterns that indicate healthy or impaired physiological status. Existing ultrasound devices for monitoring the fetal heart rate require continuous sensor adjustment as the baby moves and can only be performed intermittently under clinical supervision. As such, only limited information is known about how markers of fetal wellbeing change outside the hospital environment. We present a smart wearable device capable of continuously monitoring the fetal heart rate throughout pregnancy. This will enable new research into how environmental and maternal factors influence real-time fetal wellbeing.
Recent work by our group has demonstrated that the fetal heart rate can be reliably extracted using a fixed placement of bioelectrical sensors on the maternal abdomen. Using this sensor configuration, we have used deep learning techniques to extract extremely accurate fetal heart rate measurements when compared to traditional signal separation methods. While we shall validate the developed architecture for synthetic and physiological fetal signals, the research outcomes will have applicability across multiple domains for analysing complex multichannel data. The outcomes of the project are due to long standing collaborations with hospitals across multiple countries, clinical teams and private enterprises.
10 AM – 11 AM
|Active Target Defense with Realistic Dynamics
|Ajay Kumar Sandula
|Multi-armed Bandit Approach for Optimizing the Task Sequences of a Fixed-base Robot in a Warehouse Environment.
|Verifying Temporal Properties of RNNs
|Ashraf Haroon Rashid
|Demonstrating Adversarial Attack and Defense for Autonomous Driving Agent
|Risk-Predictive Path Planning for Human-Mobile Robot Collaborative Warehouse Environment
|Deep Unsupervised Skills Learning for Autonomous Vehicles in Urban Scenarios
|Fault Tolerant Heterogeneous Collaborative Framework for Automated Agriculture
|Automated Crowd Parameter Estimation and Crowd Movement Analysis in Kumbh Mela
|P S V S Sai Kumar
|Autonomous Soft Landing of UAVs using Minimum Jerk based Guidance Strategy
|Cyber Security of Multi-Agent Systems: Passivity-based Attack Identification and Mitigation
|Funnel-based Reachability Control of Unknown Nonlinear Systems using Gaussian Processes
|Vidyavardhaka College of Engineering
|Analysis of Various Social Media Phishing Attacks
|CORNET: A Co-Simulation Middleware for Robot Networks
|Digital-twin for large-scale distributed real-time optimization of 5G-Networks
|Consensus-based Formation Control Using Novel Signed Protocol
|A Generalized Collaboration Model for Rideshare and Transit Service Providers
11 AM – 11:45 AM
One of the major impediments in the deployment of Autonomous Driving Systems (ADS) is their safety and reliability. The primary reason for the complexity of testing ADS is that it operates in an open world characterized by its non-deterministic, high-dimensional and non-stationary nature where the actions of other actors in the system are uncontrollable from the ADS perspective. This is a state space explosion problem and one way of mitigating this is by concretizing the scope for the system under test (SUT) by testing only for a set of behavioral competencies which an ADS must demonstrate. A popular approach to testing ADS is scenario-based testing where the ADS driving stack is presented with driving scenarios from the real world (and synthetically generated) data and expected to meet defined safety criteria while navigating through the scenario.
We present SAFR-AV, an end-to-end ADS validation platform to enable scenario-based ADS testing in a data-driven manner. Our work addresses key real-world challenges of a) building an efficient large scale data ingestion and search engine to identify scenarios of interest from real world data, b) creating digital twins of the real-world scenarios to enable SIL testing in ADS simulators and, c) identifying key scenario parameter distributions to enable optimization of scenario coverage. These modules along with others such as Coverage Optimizer, Intelligent Sampler and Safety Analyzer would allow the SAFR-AV platform to provide ADS pre-certifications.
11:45 AM – 12:30 PM
This talk focuses on stochastic optimization problems defined over a network and explores data-driven distributionally robust (DR) solution methods to solve it. Specifically, we will look at chance-constrained optimization that finds application in generation planning problem and expectation minimization problem that is motivated by distributed optimization and federated learning. The DR formulations of the problem have attractive statistical guarantees but pose computational difficulties. The talk will provide algorithms to handle these challenges, paying special attention to the large-scale nature of the problem and the fact that the data about the uncertainty cannot be aggregated at one single location in the network. We will end the talk with future challenges and research directions.
12:30 PM – 2 PM
2 PM – 3:30 PM
In the last decade, there has been extensive deployment of robots in agricultural fields for various applications. In this presentation, I am going to specifically talk about a mobile sensor network developed in my lab for soybean phenotyping. I will talk about planning and estimation problems that arise in deployment of the mobile sensor network. Next, I will present a novel UAV-UGV system developed in my lab for battery refuelling. Finally, I will present planning and scheduling problems that arise in deployment of the field-robotic system for long-term autonomy.
3:30 PM – 4 PM
Human - Robot co-existence and collaboration in warehouse automation ecosystems has enabled the creation of intelligent solutions by combining human adaptivity with robotic efficiency. With a growing demand for higher warehouse throughput owing to the ever expanding e-commerce industry, expediting robot operations is becoming increasingly necessary. However, such a change in operational paradigm imposes very strict requirements on the safety of humans and other robots in a warehouse. Recent advances in low latency, high performance network infrastructure such as 5G/WiFi6 present a great opportunity to create mobile automation solutions with safety guarantees by leveraging the awareness via network. In other words, such a high quality network infrastructure opens up an additional layer of perception that is not strictly limited to on-board sensors on a robot such as lidars and cameras. This talk focuses on a few recent activities at ARTPARK-IISc that showcase the benefits of augmenting on-board sensing on robots with network-awareness. Specifically, the focus will be on a few applications such as proactive collision detection and avoidance in multi-robot systems, safety behaviors around humans and a co-simulation framework for robots and networks.
4 PM – 4:30 PM
4:30 PM – 6:30 PM
|Distributed fixed-time orientation synchronization with application to formation control
|Analysis of Inter-Event Times in Linear Systems under Event-Triggered or Self-Triggered Control
|Fixed-Time Dynamical System Approach for Solving Time-Varying Convex Optimization Problems
|Nishchal Hoysal G
|Reinforcement Learning Based Sequential Optimization for Multi-robot Intersection Management
|Worst-Case Scenario Evasive Strategies between Dubins’ Vehicles with Partial Information
|Robust Blood Glucose Regulation for Type-1 Diabetic Patients
|Vinay Krishna Sharma
|Eye-Gaze-Controlled Safe Human-Robot Interaction System for Persons with Severe Speech and Motor Impairment
|CORNET: A Co-Simulation Middleware for Robot Network