SMART SPIN COATER

PROJECT REPORT | Tinkering Lab, IIT Roorkee

By: - Ankit Kumar

OVERVIEW
Spin Coating is a standard laboratory procedure which involves depositing thin films of materials,
often polymers, uniformly on flat substrates by employing the concept of centrifugal force. The
substrate needs to be rotated at high speed to achieve the desired thickness of the coating material.
But the present day spin coaters have a limitation that they cannot control the film thickness and
texture during the thin-film coating. So, we made a device capable of controlling the
aforementioned parameters and thereby reducing the overall fabrication time of a desired thin film
by eliminating the need for solution optimization and film characterization. A four-point laser
technique was developed for the same and setup/apparatus was made using 3d-printed
components. The machining for the casing of the device was done using the hand tools and the
integration of microprocessor, four-point laser and sensors was done using the soldering facility
in the lab. The final product that we made out of this is about half the cost of existing spin-coaters
and has an added feature of real-time thickness control which could accelerate the research in
solar-cell fabrication, perovskite and organic LED fabrication and other thin-film applications.

LAB FACILITIES USED

3D Printers, Soldering Facility and Hand-Tools, Carvey.

HOW DID THE LAB HELP YOU IN YOUR PROJECT?

By using the above facilities in Tinkering laboratory,I successfully completed my project .
Figure 12: Smart spin coater
 PERISTALTIC PUMP
PROJECT REPORT | Tinkering Lab, IIT Roorkee

By: - Ankit Kumar

OVERVIEW

Peristaltic Pumps are widely used in the Pharmaceutical and Medical sector for their precise
pumping and zero-contamination applications. However, the peristaltic pumps currently available
in the Indian Market are very costly due to which, many labs cannot afford it. I modified the
pump’s gear assembly design and optimized the process of its fabrication towards an automated
process which was otherwise, very difficult to achieve using a FDM (Fused Deposition Modeling)
-type 3D Printer. For the process of optimization, the gear-assembly was 3D Printed multiple times
under different set of parameters until the best set was achieved. Moreover, the whole casing of
the device was designed here itself and was 3D Printed in ABS material which can provide strength
and flexibility using the optimized infill amount and pattern which was ‘tetrahedral’ in this case
which could further minimize the wearing of the device with time. The soldering facility in the
Tinkering Lab was also used to integrate the pump with Arduino Board to achieve precise pumping
with variable flow rate and micro-volume-dispensing capabilities. In the end it was found out that
it is possible to fabricate the device in under ₹4,000 which is normally available in the market at
₹80,000 and above. This device can be used in Hospitals, Research Laboratories, Food-Processing
Industries and in Educational Institutions like IIT’s, IISER’s and IISc and NIT’s etc. for
experimentation purposes. This would make quality research possible in numerous labs across the
nation where it is not possible using the existing technology.

LAB FACILITIES USED

3D Printers, Soldering Facility and Hand-Tools

HOW DID THE LAB HELP YOU IN YOUR PROJECT?

Lab helps a lot to print parts and other facilities – Soldering, and hand tools for my project.
Fig 13: CAD Rendering of Pump’s Gear Assembly Fig 14: Pump’s Gear Assembly (3D Printed)

Fig 15: CAD Rendering of Peristaltic Pump Fig 16: Peristaltic Pump (3D Printed)
 MECHANICAL SAFETY DEVICE FOR THREAD
FAILURE IN POWER SCREW BASED LIFTS
PROJECT REPORT | Tinkering Lab, IIT Roorkee

By: - Nitesh Arora

Prabhpal Singh Seerha

OVERVIEW

Possible Designs for Safety Device : Before discussing any safety mechanism an extra addition
has to be done on the radial drilling machine. That extra addition includes a chain that is attached
to arm to control the arm in case of failure through the responding actuator; and a sprocket and
clutch mechanism that helps in locking and stopping the motion of sprocket thus stopping the arm.
A retracting spiral spring will be attached to the sprocket to rewind the chain when the arm moves
up. Thus this additional setup is used in all the cases and methods discussed below to control the
motion of arm .The design of proposed mechanical sensor is completely new and deduced by us
for such screw based lifting devices. Assumption can be justified as if the threads are wearing out,
they will wear only because of arm’s load not by the weight of sensing nut. Moreover, the extent
of wearing required for the fall or disengaging of sensing nut is far more than required for heavy
arm. Sensing nut is provided with slots that will lock the orientation of nut. It will avoid the free
rotational motion of nut about the feed screw because of friction. If rotation is allowed, then in
normal functioning it will not move in contact (not attached) with the arm. Thus, it will ensure that
the only motion of sensing nut is up and down or vertical, in contact with arm. But when the screw
fails the arm will fall vertically and the nut will remain intact with the screw. Hence there is a
relative motion between sensing nut and arm which can be transferred to actuating part. The design
presented is novel and completely mechanical.

LAB FACILITIES USED

3D Printers- Desktop and Projet, drill, filer, and Hand-Tools

HOW DID THE LAB HELP YOU IN YOUR PROJECT?

As the previous images show that a working model was created with scrap from Team Robocon
workshop have some limitations in functioning other than aesthetics drawback. For our final
presentation to IPR committee, we needed a model that closely resembles the exact radial drill
machine with safety device i.e. same as our CAD model that we have prepared. Resembling model
gives a better understanding of the mechanism and helped us in highlighting the novelty of our
device. Further it increases the effectiveness, being scaled down model of the whole device. We
3D printed our model on Industrial 3D printer (Projet) available in Tinkering Lab, IIT Roorkee.
Industrial 3D printing made our task easy in terms of fabrication and assembling.
Fig 17 : Image showing exact replica (scaled down) of the radial
drilling machine with implementation of safety device.
 ACCELERATION-DECELERATION
CHARACTERISTICS AT ROUNDABOUTS
PROJECT REPORT | Tinkering Lab, IIT Roorkee

By: - Gaurav Kumar,

Jayesh Jain,

Pranav Anurag

OVERVIEW

We acquired the video feed of several roundabouts from the Transportation Department and then
worked on the video feed to extract the data points ourselves. These roundabouts were situated
at Chandigarh, Ibrahimpatnam and Mumbai. We conducted some statistical tests to check how
different variables are distributed and check some of the assumptions of various models. And
then, we modelled entry speeds for the vehicles turning left, right and going straight from the
roundabout. Part of the project including the statistical parts majorly was done in R and part of
the project including models majorly was done in Python. Using the algorithms selected using
cross validation, we were able to predict the Safe Entry Speed in the roundabout given various
factors like number of vehicles in circulation, number of vehicles conflicting the path of the
given vehicle.

LAB FACILITIES USED

Workstations

HOW DID THE LAB HELP YOU IN YOUR PROJECT?

By using these high configuration systems, we have smoothly working in our project.
Fig 18: Certain convention for naming the incoming legs and the weaving lengths.
 MULTIMODAL LEARNING
PROJECT REPORT | Tinkering Lab, IIT Roorkee
By: - Gaurav Kumar,
Jayesh Jain,

Pranav Anurag

OVERVIEW

The term Multimodal is used to refer to multiple kindsof data. In our case, the data
comprises of 3 different modalities -Visual Modality: The video frames of the trailers,
posters and other stills of the movie available on the internet. Audio Modality. The audio
clip of what's being spoken in the video. Textual Modality: Subtitles in the said videos,
description and other details of the movie. Multimodal learning can be grouped into 2
methods: Multimodal Regression and Multimodal Classification Multimodal Regression
is learning to predict the score of a movie. It falls in the scope of predictive modeling
which uses statistics to predict outcomes. But in our project we used deep learning models
to regress the score. Multimodal Classification includes predicting the genre of the movie
using deep learning from a list of genres. Outline as follows: Scraping a dataset, Data
preprocessing, Intuitive theory behind Deep Learning, Deep Models for Visual Data,
Deep Models for Text and Combining the different models

LAB FACILITIES USED

Workstations

HOW DID THE LAB HELP YOU IN YOUR PROJECT?

By using these high configuration systems, we had smoothly completed our project.
 INTER IIT TECH MEET-2018

PROJECT REPORT | Tinkering Lab, IIT Roorkee

By: - Vikash Kumar Meel,

Shivanshu Raj Shrivastava

OVERVIEW

This report contains detailed description of facilities used by InterIIT Tech Meet Contingent
2018 in the following projects during December 9, 2017 - January 5, 2018: Safety Device
for Fishing Vessel, Soldier Support and Warehouse Inventory Check

LAB FACILITIES USED

3D Printers, Power Tools ,3D Scanner, Electronics equipment, Welding machine and power
tools

HOW DID THE LAB HELP YOU IN YOUR PROJECT?

1) Safety Device for Fishing Vessel:3D Printing- One box of size- 160x60x100mm,
weight- were printed, Power Tools- Power saw, Dremel, Electronics- Constant voltage
power supply, Soldering iron

2) Soldier Support:3D Printing- 10 boxes of size-100x80mm, weight-20g each were

printed, 1 Major Equipment Holding for Soldiers of size-250x250mm, weight- 250 g,
3D Scanning- 3D Scanning of helmet was done.

3) Warehouse Inventory: 4 Landing Gear Legs, weight-18g each, Frame extension for
Protection Gear-4, weight-150g each were printed || Bases for Various Electronics Devices,
Total Weight -80 g (approx.), Periscope- mobile base -120g.

Special thanks to Tinkering Lab Team for their help and support.
Fig 19: Different 3D Printed parts
 NUMERICAL UNDERSTANDING OF FIRE IN
PATTERNED OBSTRUCTIONS:
EFFECT OF SPACING, MULTI-POINT IGNITION, AND
WIND SPEED
PROJECT REPORT | Tinkering Lab, IIT Roorkee

By: - Sandeep Kumar

OVERVIEW

An approach is developed to understand the propagation of flame and the factors influencing its
behavior as it spreads through the patterned flammable obstructions. We have illustrated different
stages of the fire propagation, from ignition to flash-over and ultimately to extinction because of
inability in crossing the excitation energy barrier for combustion. An analytic scale analysis
compares the several timescales for energy transfer through the fuel, resulting in a self-similar
flame affected area for different quantities of patterned flammable obstructions. Further, the Heat
Release Rate Per Unit Area (HRRPUA) of the source of fire is varied to explore a wide range of
cases with different intensities of fire ignition. Investigation of the effects of changing the packing
density of the boxes yields a power law relation for the flame affected region with good accuracy.
Further, variation in the point of ignition results in a change of the characteristic shape of the burnt
area, post-combustion. The spread of flames from multiple ignition points and their interaction
have also been investigated. At last, we have focused on the fire propagation in presence of the
wind. The centroid of the flame projected area is evaluated to determine its kinematics as
interaction happens with the different inertial strengths of blowing wind.

LAB FACILITIES USED

Workstations

HOW DID THE LAB HELP YOU IN YOUR PROJECT?

By using these high configuration systems, we have smoothly working in our project.
Fig 20: A typical fire propagation scenario: illustration of (a) flame front propagation and (b)
soot flow pattern at different temporal instances given by: (i) 1 s, (ii) 20 s, (iii) 40 s, (iv) 60 s, (v)
90 s.
 BAJA SAE INDIA 2018 | TEAM KNOX
PROJECT REPORT | Tinkering Lab, IIT Roorkee

OVERVIEW

Manufacturing of our all-terrain vehicle for participation in BAJA18.

LAB FACILITIES USED

Workstations,3D Printers, Bench drill and Hand drill.

HOW DID THE LAB HELP YOU IN YOUR PROJECT?

Workstations: Advanced computational power of the HP workstations, was utilized to run

complex analysis of various sub-parts and assemblies of the ATV. The software such as ansys,
matlab etc. were used for evaluative purposes like topology optimization, finite element
analysis, impact and roll-over simulations.

3D Printers: Prototype of the roll cage of our ATV was designed and 3D printed, to
evaluate various design related aspects. It was also used for presentation related purposes,
before actual roll cage was manufactured.

Bench drill: It was used to make holes of varying diameters, in iron sheets that varied in
thickness from 2 mm to 5 mm. The availability of this power tool, greatly reduced our
dependence on other sources, and saved a lot of time.

Hand drills: These were used to drill holes in the sheets that were less than 2 mm in
thickness.
 AUTONOMOUS QUADCOPTER CONTROL USING
REINFORCEMENT LEARNING
PROJECT REPORT | Tinkering Lab, IIT Roorkee

By: -, Akshay Shrikantrao

Anuj Jain
Yash Goel

OVERVIEW

In recent years there have been many successes of using deep representation in reinforcement
learning. It has yielded proficient controllers for complex tasks like mastering the game of go,
achieving human level performance in Atari 2600 also the development in computational power
of computer has given us opportunity to use Reinforcement learning in the field of Unmanned
Aerial Vehicle. Our work will focus on modelling and designing of control of quadcopter using
learning based control systems for stochastic real world environment. It is a particularly
challenging problem because the quadcopter is highly nonlinear and multi-variable system plus
the stochasticity in real world make this more complex and difficult to train. The problem will be
tackled using control algorithms and reinforcement learning. Quadcopter will able to navigate in
environment, detect and avoid obstacles. The robustness of our quadcopter will be demonstrated
by autonomously flying it from one location to another.

LAB FACILITIES USED

Workstations

HOW DID THE LAB HELP YOU IN YOUR PROJECT?

By using these high configuration systems, we have smoothly working in our project.
 SEMANTIC SEGMENTATION OF IMAGES USING
CONVOLUTIONAL NEURAL NETWORKS
PROJECT REPORT | Tinkering Lab, IIT Roorkee

By: -, Manpreet kaur

OVERVIEW
I am working in deep learning area, which means, to design an algorithm that works same
as a human mind. The model work according the experience learnt during training phase. The
Neural networks as named so because they work same as human’s neuron system works. As it
need a large amount of training data (in this project the data is in the form of image), so it need
a big part of system’s memory. And since we need a model which can work in real time system,
real time mean the system which can work quickly on the input data provided, so in order to
achieve a system with high computation speed is needed. In this project, I am working on Linux
Environment. The language in which I am working is python.There are ground truth values given
in the data set. Correspondingly image annotations are given. Annotations includes the classes
of the objects appearing in the image. Firstly, I have to over-segment the image in-order to divide
the image into super-pixels. For this I have used the SLIC (simple linear iterative clustering).
SLIC being the fastest algorithm among all segmentation algorithms takes less time at initial
stage of the project. I completed this step. Our CNN model works on this over-segmented image.
Rather than working on the pixels, model take less time on super-pixels, thus giving output in
lesser time. The model consists of different layers: Convolution layer, Max-pool layer, RELU
(rectified linear unit) layer and deconvolution layers. This model is producing the feature map.

LAB FACILITIES USED

Workstations

HOW DID THE LAB HELP YOU IN YOUR PROJECT?

By using high configuration systems, I had smoothly worked. Also the staff is very helpful. We
got our all issues sorted within time. And all facilities were provided with no restriction that were
required to do our work.
 ACTION RECOGNITION USING FEATURE FUSION
PROJECT REPORT | Tinkering Lab, IIT Roorkee

By: -, Renuka sharma

OVERVIEW

Human action recognition is an important area in the field of computer vision due to increasing
security and day to day concerns. Nearly all the shops, malls, stations and institutions have
Closed Circuit Television (CCTV) cameras installed so that they can keep track of human
activities and identify the suspicious activities before something bad happens. Nowadays the
technology is advancing so much that one day, in the near future, all this will be tracked by the
intelligent systems and the work is going on in all the fields to enhance the capabilities of the
intelligent systems that will be able to assist us and ease our work. So, human action recognition
is also an important area of research and here we are finding some more efficient ideas than the
existing ones to get there. The reason being, we need intelligent systems to observe the videos
for visual surveillance, human-machine interfaces, sports video analysis and video retrieval in
the best way possible. The human action is a sequence of body movements which varies from
the sim-plest movement of a limb to complex joint movement of a group of limbs and body.In
order to identify the action of the human being, we need to observe the movement and then
match it with the previously defined patterns and then assign it a label.

LAB FACILITIES USED

Workstations

HOW DID THE LAB HELP YOU IN YOUR PROJECT?

I have been working in the Tinkering lab since past 6 months now and it has been great help for
me as I was able to get the system with required configuration so that a Deep learning based
project could be carried out. The system that I am working on has 32GB RAM and enough
space to keep all my data. Lab’s coordinators have always been helpful enough to help me deal
with the problems that came my way while I was doing my project. I sincerely thank all of the
staff who are behind the Tinkering Lab’s proper functioning and hope that other students will
also get benefited from it.