1.
AI driven engineering drawing dimension identification and parts & Dimensions correlation
system
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
In the field of engineering and manufacturing, technical drawings play a crucial role in conveying
detailed information about a product or a part. These drawings include multiple dimensions that are
critical to the product’s assembly, functionality, and quality. However, interpreting these dimensions
and understanding their interrelationships can be complex.
The challenge lies in developing a method or a system that can accurately identify and understand
the various dimensions presented on a drawing sheet. This includes recognizing the dimensions,
establishing meaningful connections between them, and understanding how they relate to the
overall part or product. The ultimate goal is to derive a correlation between the type of part and its
associated dimensions, which can provide valuable insights for manufacturing, quality control, and
product development. This could potentially lead to more efficient conversion, improved product
quality, and significant cost savings.
Challenge / Requirement
Develop a method or system that can accurately interpret engineering drawings. This system should
be capable of identifying various dimensions on a drawing sheet, understanding their
interrelationships, and correlating them with the type of part being represented. The ultimate goal is
to enhance the efficiency of manufacturing processes, improve product quality, and achieve cost
savings. This challenge requires a deep understanding of technical drawings and the ability to
translate this understanding into a practical, automated solution. It’s a complex task that involves
elements of computer vision, machine learning, and mechanical engineering.
Deliverables
A functional prototype that can accurately interpret engineering drawings, identify various
dimensions, understand their interrelationships, and correlate them with the type of part being
represented.
2. Developing a wind regenerative system for EV charging using air-pressure
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
Electric Vehicles (EVs) currently lack a wind regenerative charging system that utilizes air pressure.
This means that while EVs can recapture energy through braking, they are not yet able to harness the
energy potential of air pressure during the vehicle's motion.
Challenge / Requirement
Develop a battery charger that employs air resistance and an electric motor to generate voltage. This
system should support the low-voltage functions in EVs and be complemented by a solar system
integrated into the vehicle's roof. The combination of these charging methods aims to achieve the
necessary voltage to efficiently charge the EV battery.
Deliverables
A proof-of-concept demonstration, along with test results showing the efficiency of the charger in
various conditions and its integration with the solar system
3. AI Powered Smart Component Management
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
In the field of Component Engineering, a crucial task is to identify electronic components that have
reached their End of Life (EOL) using tools like IHS or Silicon Expert. After pinpointing these EOL
components, the process of finding Form, Fit, and Function (FFF) replacement components is usually
performed manually. This involves extensive research on websites and analysis of component
datasheets.
Challenge / Requirement
Automate the manual process with an intelligent AI system. A system would be capable of identifying
EOL components and suggesting alternative FFF components without human intervention.
Deliverables
AI engine which takes input as component details and finds replacement component automatically
4. Plug n Play Tool for battery liquid cooling plate design
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
The design of a battery's liquid cooling plate is crucial for managing the thermal load. The current
design process is not efficient, requiring numerous iterations, significant computational resources,
and extended timeframes, which can delay the overall development process.
Challenge / Requirement
Develop an optimization tool that can adapt swiftly to changes in product design, including
alterations in shape and size, without compromising the delivery schedule or inflating the design
costs. The tool should seamlessly integrate with the existing infrastructure to improve the battery
cooling plate's performance, ensuring uniform velocity and temperature distribution.
Additionally, the goal is to develop a design approach for the liquid cooling plate that can reduce the
required pumping power by 30%.
Deliverables
Thermal performance prediction (velocity, temperature and pressure drop) by Varying battery
cooling plate dimensions, patterns, sizes and shapes
5. Vehicle underbody inspection system
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
Modern vehicles are increasingly equipped with 360-degree cameras to provide a comprehensive
view of the surroundings. However, the underbody area remains a blind spot, posing a risk of
accidents involving objects, animals, or humans that may be underneath a vehicle during parking or
slow movements. A system that enables visualization of this area would significantly enhance safety
by eliminating the need for physical checks.
Challenge / Requirement
Develop a system that allows for underbody visualization under various harsh conditions, which can
be integrated into the vehicle's cabin display. This system should be adaptable to various vehicle
types and potentially connect to the driver's mobile device, providing a seamless and user-friendly
interface. Good to get integrated with ADAS features.
Deliverables
Working concept
Tentative BOM
Implementation Plan
Budgetary cost calculation
Regulations assessment
6. Enhanced Cancer treatment outcome through Simulations
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
'The application of electromagnetic fields in cancer treatment represents a cutting-edge approach
that prioritizes patient safety through minimally invasive procedures. This technology necessitates
the creation of a virtual environment to simulate surgeries and predict outcomes with high accuracy,
ensuring effective and safe treatment plans for organs such as the heart, kidneys, lungs, bones, or
liver.
Challenge / Requirement
'REfine the catheter design to minimize thermal damage, utilizing virtual models for real-time
visualization of the ablation zone, and delivering precise electromagnetic energy to target only
malignant tissues. Develop a robust design that incorporates dynamic effects from the physical world
and evaluates various probe structures for their impact on different cancer cell shapes.
Deliverables
One can predict the dose requirement precisely based on how much energy will be needed for a
typical tissue of interest as per its malignancy and in real time any complex scenario can be built in
simulation environment.
7. AI powered Neo-natal Vital monitoring
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
In neonatal intensive care units (NICU), constant monitoring of infants' vital signs is crucial for their
survival and health. Traditionally, this monitoring requires a medical attendant to be physically
present, which can be resource-intensive. An example of the importance of such monitoring is the
detection of a drop in SpO2 levels, which indicates a need for immediate oxygen therapy to prevent
serious complications.
Challenge / Requirement
Develop an AI-based monitoring system for neonates in the ICU. This system would track not only the
vital signs but also the sleep and activity patterns of the infants. Such a system would enable prompt
medical intervention, allowing for the immediate response to emergencies, early detection of health
abnormalities, and more effective treatment.
Deliverables
Develop application which records infant vital signs like respiratory rate (RR), heart rate (HR),
temperature (T), blood pressure (BP), and oxygen saturation level (SpO2). Record Infant Active/Sleep
patterns and inform on any abnormalities.
8. Multicast over Cloud for IPCCTV
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
As smart cities grow, so does the use of cameras for surveillance, leading to a high demand for live
streaming. This streaming requires substantial bandwidth, which is a limited resource. Multicasting
can help by allowing multiple users to view the same stream simultaneously, but it's not widely
supported by public cloud providers due to security concerns, although it is available in on-premises
data centers.
Challenge / Requirement
Develop a system that allows multicast video streams from IP cameras to be sent both to a command
center and a cloud data center at the same time. This system must ensure that cloud data centers, or
Hyperscalers, can receive these multicast streams on their virtual machines (VMs) and use them for
Video Management System (VMS) applications, all while maintaining security and operational
integrity.
Deliverables
A solution note comprising of deployment architecture and proof of concept
9. 5G network components Interface verification approach
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
The 5G ORAN Architecture introduces a disaggregated network structure, where multiple interfaces
such as E2, O1, O2, F1, N2, and N3 are defined, each with its own set of Key Performance Indicators
(KPIs). This complexity makes it challenging to develop a unified and straightforward method for
interface verification.
Challenge / Requirement
Create a simpler and standardized approach to verify the various interfaces of 5G network
components, ensuring consistency and reliability across the board. The approach can use any
opensource code to build the solution. The approach need to be as per ORAN standard.
Deliverables
A simpler and standard verification approach for 5G network components Interfaces.
10. An approach for Simpler 5G UE simulator (Open source stack) for basic 5G network testing
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
5G technology brings a transformative shift in wireless communication, offering significantly higher
speeds and more reliable connections for User Equipment (UE). However, as 5G is still in its nascent
stages, the simultaneous development of its components presents a substantial challenge in testing
and validation. Ensuring that all parts work harmoniously is crucial for the successful deployment of
5G networks.
Challenge / Requirement
Targeting the complexities of testing in 5G networks, develop a simplified 5G UE simulator using an
open-source stack is proposed. This simulator, coupled with Software Defined Radios (SDRs), must
aims to facilitate basic network testing. The design to be a cost-effective solution that can also
support performance testing, providing a practical tool for developers and researchers.
Deliverables
An approach for Simpler 5G UE simulator (Open source stack) for basic 5G network testing. UE Sim
with SDRs. Can be used for perf testing as well.
11. Multi-metal 3D printing for complex components
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
Industries becoming more cost oriented and need of supplying very highly customized product. A
unique solution of building the Major Metal assemblies using 3D Printing technology is evolving. But
Multi-material (combination Materials like in an assembly) is still a challenge to produce. A PoC or
concept to make it practically is an area of exploration in current world.
Challenge / Requirement
Develop a 3D printing process capable of handling multiple metal materials simultaneously. The key
objective is to create a method that allows materials such as aluminium, steel, and copper to be
printed either sequentially or in parallel, while maintaining their individual melting points and
physical integrity. A successful solution must ensure that the fusion of materials does not
compromise their properties, although some trade-offs in print resolution may be acceptable.
Deliverables
A prototype or simulation demonstrating the multi-metal printing capability.
12. Latent Emotion detection of a machine
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
'Latent sentiment or emotions refer to the subtle emotional undertones conveyed in text without
direct expression through explicit words. It's the hidden emotional layer that can be discerned from
the context and the choice of words. For instance, when a machine's function includes an "occasional
oil dip," it may suggest a maintenance issue, even if not stated outright. Similarly, a sound level of
"150 dcb" could indicate a problem due to excessive noise, and an "Error, process halted" message
clearly points to a critical issue. The absence of a "Try catch statement" in Python code implies a
potential risk for unhandled exceptions.
Challenge / Requirement
The challenge lies in identifying these latent sentiments. Solution must learn to read between the
lines, understanding that language is not just about what is said, but also what is implied. For
example, the phrase "Machine function, occasional oil dip" subtly hints at a need for attention, while
"Machine function, occasional it sounds 150 dcb" suggests a significant issue with noise levels.
"Machine state: Error, process halted" signals a negative situation, and the lack of a "Try catch
statement" in Python code indicates a missing layer of error handling.
Deliverables
May use inter discipline subject principles with AI / Gen AI and expectation is to define the Emotion
behind the phrase e.g; Machine function, occasional oil dip
Emotion: Negative Emotion as it's unusual behavior
13. Improving Explainability of Medical Diagnosis Systems
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
Medical diagnosis systems are increasingly being used to aid healthcare providers in making
decisions. However, these Deep learning systems are often seen as "black boxes" that lack
transparency and interpretability, making it challenging to understand how they reach their
conclusions. This lack of transparency can lead to mistrust and impede the adoption of these systems
in healthcare, where accountability and transparency are crucial.
Challenge / Requirement
The challenge is to develop methodologies that enhance the explainability of DL based medical
diagnosis systems without compromising their performance or utility. The goal is to improve
transparency, interpretability, and trust in these systems so that healthcare providers can make
informed decisions based on the system's output
Deliverables
1. A practical framework or toolkit for integrating explainability into DL based medical diagnosis
system development, ensuring that it is a fundamental aspect of system design rather than an
afterthought.
2.The demo should cover various approaches to improving transparency and describe their strengths
and weaknesses.
14. Developing innovative solutiom for chronic Insomnia
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
'Sleep disorder is one of prime health ailments which is significantly impacting the working
population around the world. Research indicates that approximately 8% of working adults suffer
from chronic insomnia 1. Additionally, sleep apnea affects nearly 1 billion people worldwide, many of
whom are part of the working population. These disorders can lead to reduced productivity,
increased absenteeism, and presenteeism, resulting in substantial economic costs.
Challenge / Requirement
1. Develop a sleep therapy method to help patients to fall asleep without significant delay and
remain asleep for a necessary amount of time to feel rested.
2. Find a method to integrate the sleep therapy method with the sleep diagnostic and monitoring
instruments, so that therapy can be applied based on the monitoring parameters.
Deliverables
A product/ Instrument/ Device/ Mechanism that can meet the challenge statement.
15. Workforce efficiency in Factories
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
'Efficient factory worker management is the key to operational excellence in numerous industries,
including manufacturing, food processing, textiles, and pharmaceuticals, among others. Monitoring
workforce efficiency in factories is a challenge for the management. Optimizing labor deployment to
adjust staff levels to match production needs throughout the day to maximize output. For example,
adjust staffing levels based on the product being run on the line or let operators call it a day if
production ends early.
Challenge / Requirement
Develop an AI-driven system that employs computer vision to oversee worker movements in real-
time across the shop floor, especially near hazardous machinery and designated pathways. This
system should facilitate
- live headcounts
- workforce efficiency by tracking idle workers,
- managing crowd /queue management through alerts when a zone exceeds required worker count
- task-specific workforce monitoring including duration,
- optimizing labor deployment based on skills and productivity for critical tasks
- enhancing security by detecting trespassing and loitering, with immediate alerts for any breaches
Deliverables
1. Computer vision based system for image processing.
2. AI Model for work force efficiency
3. Middleware to process the AI Model output and interface to cloud for monitoring.
16. Drive by wire simulation on a HIL bench with drive visualization display
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
Often needed to simulate the vehicle driving control in vehicle system software design cycle.
It is tedious and complex to test each vehicle software change on an actual vehicle. In house
simulation of software on a test bench saves time and cost of development of vehicle solutions.
Challenge / Requirement
Develop a test banch that involves setting up a HIL based vehicle control actuations of throttle,
steering and braking actuation signal simulation.
As part of the test bench, on applying brakes or throttle or steering the driving wheel, the vehicle
control system sends a CAN format message to the Actuation system to convey the braking,
accelerating/decelerating or steering action. On receipt of such CAN messages, the actuation system
must generate the actuation signals connector to brake, steering wheel & throttle actuators. Road to
be sensed from steering wheel. Additionally, as and when such actuations take place, the vehicle
dashboard visualization can come on display monitor screen to show simulate vehicle forward
motion or speed reduce, steering and braking using a platform such as CARLA.
Deliverables
Demonstrate a fully functional simulation code that manages vehicle control actuations for throttle,
steering, and braking. The code should be able to interpret CAN messages and translate them into
actuation signals. Additionally, provide a visual dashboard interface that reflects the vehicle's actions
in real-time, using a simulation platform like CARLA. This interface must accurately display the
vehicle's motion, speed variations, and responses to driver inputs such as braking and steering.
17. Development of a Robotic System for Wind Turbine Bolt Inspection and Maintenance
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
Wind Turbine Tower sections are super heavy structures joined together to form a tower on top of
foundation and anchor cage. These sections are bolted together using huge bolts, some as large as
M75, to ensure everything stays in place despite the forces of nature and wear and tear like rust.
These turbines undergo regular manual check-ups which is expensive and risky.
Challenge / Requirement
Create a robotic system capable of routinely inspecting and securing every nut on the tower's
sections. The robot should be able to move efficiently between sections via platforms autonomously
navigating both around and up the tower sections while ensuring transition from one area to another
Deliverables
A functional prototype demonstrating the robot's ability to move across sections and perform nut
inspection and tightening tasks is expected along with data logging and ground connectivity.
18. Advanced Lightening Detection System for a Wind Turbine
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
Wind turbines, as tall and isolated structures, are highly susceptible to lightning damage, leading to
significant economic losses, safety hazards, and operational disruptions. This challenge seeks
innovative solutions for early lightning detection, real-time warning systems, and robust lightning
protection measures to enhance wind turbine reliability and resilience.
Challenge / Requirement
Develop advanced technologies for early detection and effective protection of wind turbines against
lightning strikes. The cost-effective solution should be Innovative with easy to manufacture, install,
maintain et all
Deliverables
A cost-effective prototype of the lightning detection system.
19. AI based optimal Wind Turbine Site Selection
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
The selection of suitable sites for wind turbine installation is critical for the successful development
of wind energy projects. Factors such as wind speed, terrain, land use, grid connectivity,
environmental impact, and economic considerations play a crucial role in determining the viability of
a site. Traditional site selection methods are often time-consuming, costly, and rely heavily on expert
judgment.
Challenge / Requirement
Develop an AI-driven platform to streamline the wind turbine site selection process by efficiently
analyzing vast amounts of data, including meteorological, geographical, and regulatory information.
The platform should identify optimal sites based on multiple criteria, such as energy yield, cost-
effectiveness, and environmental impact. This tool needs to collect the data related to the wind
classes, wind zones, geography, grid connectivity and energy storage. By Using data analytics this raw
data should be converted into actionable insights and trends.
Deliverables
A prototype would include a user-friendly interface that displays potential wind turbine locations
based on the analyzed data.
- An AI model capable of analyzing and processing data to identify potential sites.
- A comprehensive dataset of relevant parameters (wind speed, terrain, land use, grid infrastructure,
environmental factors, etc.)
- A user-friendly interface to visualize and interact with the results.
20. CT Image Denoising
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
'In the realm of medical imaging, specifically CT scans, you might encounter two common types of
disturbances: Poisson noise, which arises due to the statistical nature of photon detection, and
periodic noise, which is often due to the interference of electronic and mechanical components in
the scanner. Understanding and mitigating these noises are crucial for clearer images that can lead to
more accurate diagnoses.
Challenge / Requirement
To create a software tool that can interpret and display CT scan images of the brain, specifically in the
DICOM format. Additionally, you're challenged to design an algorithm that can effectively reduce two
common types of noise in these images—periodic and Poisson—without compromising the valuable
data they contain. This will enhance the clarity of the images and aid in accurate diagnosis.
Deliverables
Demo of input image (CT Series of brain) display and corresponding denoised output.
SNR details of input and output.
21. Precision Blade edge detection
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
'The manufacturing of shaving blades requires stringent quality control measures to ensure product
safety and performance. Precise edge detection and measurement are critical for maintaining blade
sharpness and preventing defects. The current process is microscopic edge damage check where the
operator manually checks and records the quality data and take action if needed.
Challenge / Requirement
Design and construct a device that leverages cutting-edge technology to automate the detection of
edge discontinuities and damages in blades. The device should be capable of analyzing 24 uncoated
blades placed on a fixture with the simple press of a button, identifying and quantifying any
imperfections along the blade edges. This project aims to innovate beyond the current manual
microscope-based inspection methods, enhancing efficiency and precision in defect detection.
Deliverables
1. Go/No Go Criteria:
A technology to capture all the critical & major blade defects - technology must be capable of
capturing the defects and clear distinction to be done by inspecting the images.
2. Software:
Be able to detect any discontinuity in the edge from images of edge captured by a technology
Be able to measure linear dimensions of discontinuity or defect from images of edge captured
Be able to measure angular variation of discontinuity or defect from images of edge captured
Have an option for an operator to login with their credentials.
3. Data Visualization:
Inspection data should directly go into database and available for analysis of defects.
4. Data output required for display and storage:
Visuals of entire blade with marked edge damage areas
Number of rejects in one edge, i.e., how many sections of edge during detection got rejected.
Length of rejects in edge, laterally and vertically.
Total number of rejected blades in the set of 24 naked blades and highlight which among the 24
blades is rejected.
A PoC demonstrating the design, the technology used, and the operation of the device. It should
include schematics, code, and algorithms that drive the automation.
22. Generating Realistic Synthetic Data for ML
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
'Creating numerical data, it's not as straightforward as generating text or images because the
numbers must make statistical sense. The challenge is that the existing methods might not be up to
par for producing numerical datasets that hold up to rigorous statistical standards.
Creating numerical data, it's not as straightforward as generating text or images because the
numbers must make statistical sense. The current available current methods may not be sufficient to
generate statistically relevant numerical data.
Challenge / Requirement
Design a system that utilizes generative AI to produce synthetic numerical datasets, which can be
employed by machine learning models, based on a limited set of real data. The system should also
incorporate a mechanism to assess the quality of the synthetic data generated.
Deliverables
A fully functional prototype of the AI-based system that can generate and evaluate synthetic
numerical data.
23. AI powered content moderation
Nature of Challenge
Product Related
Domain / Discipline of Challenge
Tech
Description (With Technical Specifications)
Background
'In the media and video industry, managing content is a significant hurdle. Various platforms
distribute videos from businesses and individuals, which are available globally to viewers of all ages.
Occasionally, these videos may contain unsuitable material or breach copyright laws. Content
moderation involves both automated and manual methods due to the vast volume of videos that
require review. The sheer scale of data makes it unfeasible to rely solely on human moderation.
Continuous video uploads necessitate automated systems to prevent delays in flagging objectionable
content.
Challenge / Requirement
Create a content moderation system capable of real-time video analysis. It must effectively detect
and address any unsuitable or restricted content within videos, taking into account the vast and
unending stream of uploads. It's essential to harmonize automated precision with the discernment
of human oversight, all while adhering to the complex web of international copyright regulations and
content standards for diverse audiences worldwide.
Deliverables
A prototype that should be demonstrable, showcasing its ability to analyze and flag content in real-
time, with an emphasis on the balance between automation and human review.
