Video Scripting

Semifinal Submission

Final Submission

Introduction - 1 Min

Samvedan ID with team mates/photos/name given in PPT

Brief Intro - 1 Min

Intro about Institute - 1 Minute

DITE college introduction + IIF introduction

Project Brief - 1 Min

Innovation

as mentioned in the PPT | compare to present solutions

Application

different industries → minor and major industries

Done till now? - 1 Min

Hardware developed

rapid prototyping

modules developed

data set creation

transfer learning

data training

deployment

data acquistion

testing on pi

integration of software and hardware

Challenges Faced - 1 Min

same as PPT

Target Market - 1 Min

Who will be your competitors big companies like cognex, omron, keyence and other robotics companies who are already implementing industrial automation

what will be the costing if industrial standards are included costing will be under 1 lakh per module/application

USP - 1 Min

Edge ML - industrial environment integrate and not dependent on internet

Demonstration - 3-5 Min

Live-demo

image classification

object detection

microphone noise detection

temp sensor and GNSS data collection

MAX LIMIT - 12 Min

Preferred LIMIT - 10 Min

Cinematics Shots

→ Spresense

Slow pan shots with covering modules, submodules

showcase placement of camera in POV

integration of modules

→ Microphone showcase

showcase light according to situation

re-use previous audio from old video

re-create situation in post

show edgeimpulse platform

show audio spectrum

→ Image Classification

show RAW data

show POV

show comparable industrial situations

show edgeimpulse platform

show live demo with light indicator

light indicator → yellow : idle, green: ok, red: alert

show GUI/code on monitor

show at least 2 items in range

→ Object classification

show RAW data

show POV

show comparable industrial situations

show edgeimpulse platform

show live demo with light indicator

light indicator → green : ok, red: spaggeti

show GUI/Code on monitor

→ Temperature sensor and GNSS

show temp data on thingsspeak

show GNSS data on thingsspeak

show data acquisition

PPT

→ Overview

Included components

Hardware overview + add on

Visual indicator

hardware testing

hardware demo

software overview

software testing

software demo

→ Advantages

Edge ML

Data Analytics

Transfer Learning

Plug and Play

Modern turn key solution

safe machine to human interaction

cross compatibility

Add-on to present solutions

community building meets industrial standards

→ Challenges

lack of wireless connectivity

little information on internet

chip shortage

no access to industrial environment

→ Visual Illustrations

product mockup

flowcharts

technology used logos

hardware illustrations

solid works renders

→ Photos

All high resolution photos

spresense module

add-on modules

lighting images

visual indicator

demo images + video

Voice-Over

→ Intro

Team Mechatronics

Anshuman Fauzdar, Btech mechatronics 4th year

Sunny Vedwal, Btech mechatronics 4th year

Vinay Dhiman, Btech mechatronics 4th year

→ Institute

Delhi Institute of tool engineering, now known as DSEU

DITE innovation and incubation foundation support for startups

→ Innovation

Edge ML - tinyML models working on the microcontroller and taking decisions and data analytics without any need of internet access, cloud solution or task heavy devices.

Data Analysis - acquiring data from different sensors and devices providing suitable data points to develop a predictive model which gets accurate with increasing data flow.

Transfer learning - deployment of solution with very minimum new data required which is possible with transfer learning and implementing with tinyML models. This technique uses pre-built data points and efficient usage of new data.

→ Application

Self diagnosis/self healing - Machine taking decisions and maintaining itself by data points classified according to the situation, this will save time from unwanted breakdown or maintenance work and which can be verified by humans remotely.

Human Machine Interaction - situation where machine alerts the human and halts the process and further maintenance work is done by the operator with suitable inputs.

Consider a company where multiple machines are installed and requires constant monitoring, predictive maintenance can monitor in real time and provide transparent data which is suitable for continuous improvement and verified by expert supervision

→ Work done till now

Proof of concept developed and prototype tested

3D printed camera module ready

Modern extrusion mount ready

closest industrial machine, 3D printer as test machine

visual indicator ready

wifi + temperature sensor integrated

testing and data acquisition done

data sets ready

continuous integration and deployment tested

data set tested

model validated

custom integration done

contributions on forums done

in house developed documentation for easy communication

software and hardware integrated

→ Challenges Faced

Lack of wireless protocol such as wifi, bluetooth caused in hindrance in IoT testing

Sony have great documentation and SDK support but very less community support is available right now

due to chip shortage, there was limited choice in creating add-ons and we used existing solutions to integrate with spresense

limited access to industrial environment, due to covid restrictions and closest industrial machine we tested our proof of concept is a 3D printer.

→ Target Market

small scale industries which are not using industrial automation

dangerous industries which include energy, construction, chemical, mining

industries where human-to-machine interaction is direct and thus can cause accidents

industries where maintenance cost money to the manufacturer and consumer

→ Competitors

companies which are providing integrated industrial automation solutions and rapidly working on products with similar applications

→ Costing

Sony spresense bundle costs 10k, modular industrial hardware 50k, other industrial standards will cost around 40l

Final cost will be 1 Lakh INR per solution

→ USP

ML on edge - modules integrated without any requirement of internet or cloud solution

Data analytics - improvement of predictive model with increasing data points

Transfer learning - deployment of solution with very minimum requirement of initial data

Plug-n-play - Easy and integrated solutions for industries without any extra hassle

turn key solution - solution ranging from rapid deployment to full scale maintenance

safe human to machine interaction - eliminating accidents caused due to machines and ultimately lowering down risk of operators working with machines.

cross compatibility - extended solution in case of pre-installed solutions or hybrid requirement of application or customer

community building meets industrial standards - a new insight to industrial automation as a learn and apply approach with robust community support

→ Demonstration

Image Classification here, extrusion head is target object, unwanted human interaction with machine notifies in real time, same works for any other hindrance visual indicator can be extended to other notification methods allowing fast decisions ML model is validated with testing new data and transfer learning

Object Detection when machine gives wrong output, in this case spaghetti crosses a certain threshold, it alerts the user which allows fast decisions object detection model is validated with extensive testing

Noise Detection background noise, fan sound, and fan noise data is collected and validated for labelled data which predicts faulty fan

Controlled Lighting There are certain situations where lighting is important, our POC is working accurately in both dim and bright situations which allows dynamic applications and maintenance in real time

Sensor Data temperature, data, altitude and GNSS data is collected in real time to improve predictive maintenance models

Final Submission

→ Project Briefing

Problem Statement

Project briefing

→ Project Execution

Testing on real-life conditions

Use previous video of demonstration

Extended notification methods

CI/CD more detail in voice over

→ Technology and Innovation

Industry 4.0 → edgeML on microcontrollers - low cost, low power, no internet connection required and data stays on device providing privacy and making deployment easy and fast

Data Analytics → Process and filter data → Extracting features → Inspect hardware status, sensor data, trigger actions → Analyze → Predict → Maintain

Transfer learning → ML model deals with variations

Inventory management

Safe working space

→ Transformation

Compatibility

Conventional to smart

Industry 4.0 + Industrial IoT

Benefits

Save cost on repair of broken equipment

Shorter outage

Optimized use of maintenance staff

Optimized spare part stocking

Hybrid solutions

No hardware modification/tampering → No warranty issues

→ Knowledge sharing

Forums participation

Documentation

→ Industries

Small scale industries

Hazardous Industries

RUL

Data driven

Saving human life

→ Product Pricing

Hardware as a service

Subscription Model

Platform as a service

Final Submission Scripting

Script.txt

11.0KB

1 00:00:01,360 --> 00:00:05,159 samvedan 2021

2 00:00:06,240 --> 00:00:07,919 easy chair id

3 00:00:07,919 --> 00:00:10,320 35

4 00:00:10,320 --> 00:00:13,840 team mechatronics

5 00:00:13,920 --> 00:00:15,679 Anshuman Fauzdar

6 00:00:15,679 --> 00:00:19,320 btech mechatronics

7 00:00:20,320 --> 00:00:21,920 sunny vedwal

8 00:00:21,920 --> 00:00:25,519 btech mechatronics

9 00:00:25,599 --> 00:00:30,599 vinay dhiman btech mechatronics

10 00:00:30,800 --> 00:00:33,120 problem statement

11 00:00:33,120 --> 00:00:35,760 a way to minimize both failures and

12 00:00:35,760 --> 00:00:38,320 maintenance work while maximizing

13 00:00:38,320 --> 00:00:41,440 machinery and component life

14 00:00:41,440 --> 00:00:44,000 solution

15 00:00:44,000 --> 00:00:45,920 multi-machine monitoring on the

16 00:00:45,920 --> 00:00:48,879 production line using cameras multiple

17 00:00:48,879 --> 00:00:52,000 microphones and sensor inputs

18 00:00:52,000 --> 00:00:54,480 ai enabled decision making before

19 00:00:54,480 --> 00:00:56,879 breakdown occurs providing maximum

20 00:00:56,879 --> 00:00:59,440 remaining useful life managing inventory

21 00:00:59,440 --> 00:01:04,239 and benefiting men machine and product

22 00:01:04,239 --> 00:01:06,880 project execution

23 00:01:06,880 --> 00:01:09,280 demonstration of conventional machine

24 00:01:09,280 --> 00:01:12,920 performing tasks

25 00:01:13,920 --> 00:01:16,240 interacting with the machine without

26 00:01:16,240 --> 00:01:17,759 safety measures

27 00:01:17,759 --> 00:01:21,040 and once there is a unwanted interaction

28 00:01:21,040 --> 00:01:25,560 it notifies in real time

29 00:01:27,520 --> 00:01:30,159 here extrusion head is target object

30 00:01:30,159 --> 00:01:33,040 unwanted human interaction with machine

31 00:01:33,040 --> 00:01:36,479 notifies in real time same works for any

32 00:01:36,479 --> 00:01:39,040 other hindrance, visual indicator can be

33 00:01:39,040 --> 00:01:41,360 extended to other notification methods

34 00:01:41,360 --> 00:01:44,320 such as message mail or sound indicator

35 00:01:44,320 --> 00:01:46,560 allowing fast decisions transfer

36 00:01:46,560 --> 00:01:49,600 learning model is validated with testing

37 00:01:49,600 --> 00:01:52,840 new data

38 00:01:58,960 --> 00:02:02,000 when machine produces wrong output in

39 00:02:02,000 --> 00:02:04,399 this case spaghetti crosses a certain

40 00:02:04,399 --> 00:02:09,399 threshold it alerts the user

41 00:02:17,120 --> 00:02:19,920 object detection models are validated

42 00:02:19,920 --> 00:02:23,840 with extensive testing

43 00:02:34,800 --> 00:02:37,840 background noise

44 00:02:38,400 --> 00:02:40,080 fan sound

45 00:02:40,080 --> 00:02:42,319 and fan noise

46 00:02:42,319 --> 00:02:45,360 is collected and validated for labeled

47 00:02:45,360 --> 00:02:48,640 data which predicts faulty fan similar

48 00:02:48,640 --> 00:02:51,200 application extends to industrial

49 00:02:51,200 --> 00:02:53,680 machines

50 00:02:59,120 --> 00:03:02,879 temperature altitude and gnss data is

51 00:03:02,879 --> 00:03:05,519 collected in real time to improve

52 00:03:05,519 --> 00:03:09,120 predictive maintenance data validations

53 00:03:09,120 --> 00:03:12,080 continuous integration and deployment is

54 00:03:12,080 --> 00:03:14,879 integrated to test in real time and make

55 00:03:14,879 --> 00:03:19,040 deployment easy and fast technological

56 00:03:19,040 --> 00:03:20,400 innovation

57 00:03:20,400 --> 00:03:23,840 industry 4.0 implementing edgeML on

58 00:03:23,840 --> 00:03:26,319 microcontrollers making solution low

59 00:03:26,319 --> 00:03:28,560 cost low power and no internet

60 00:03:28,560 --> 00:03:30,879 connection required and data stays on

61 00:03:30,879 --> 00:03:33,440 device providing privacy and making

62 00:03:33,440 --> 00:03:37,200 deployment easy and fast data analytics

63 00:03:37,200 --> 00:03:40,080 using data analysis to process and

64 00:03:40,080 --> 00:03:42,560 filter data which helps in extracting

65 00:03:42,560 --> 00:03:45,519 features data including hardware status

66 00:03:45,519 --> 00:03:47,840 sensor data and trigger actions which

67 00:03:47,840 --> 00:03:51,280 provide a executable structure transfer

68 00:03:51,280 --> 00:03:54,239 learning deployment of solution with

69 00:03:54,239 --> 00:03:57,599 very minimum requirement of initial data

70 00:03:57,599 --> 00:04:00,159 machine learning models deals with

71 00:04:00,159 --> 00:04:01,439 variations

72 00:04:01,439 --> 00:04:04,080 and rapid plug-and-play deployment can

73 00:04:04,080 --> 00:04:05,920 be done

74 00:04:05,920 --> 00:04:08,159 inventory management proactive

75 00:04:08,159 --> 00:04:11,360 maintenance provides real-time data for

76 00:04:11,360 --> 00:04:14,879 maximizing use of remaining useful life

77 00:04:14,879 --> 00:04:17,358 which helps in maintaining inventory of

78 00:04:17,358 --> 00:04:21,120 replacements and raw material

79 00:04:21,120 --> 00:04:23,199 safe working space

80 00:04:23,199 --> 00:04:26,639 human to machine interaction is safe as

81 00:04:26,639 --> 00:04:29,680 feedback is in real time using extended

82 00:04:29,680 --> 00:04:32,320 notification methods and self-decision

83 00:04:32,320 --> 00:04:35,280 making techniques

84 00:04:36,320 --> 00:04:39,120 transformation

85 00:04:40,720 --> 00:04:44,400 compatible with latest machines systems

86 00:04:44,400 --> 00:04:47,360 and protocols

87 00:04:50,880 --> 00:04:53,680 upgrading conventional machines into

88 00:04:53,680 --> 00:04:57,840 smart and safe machines

89 00:04:59,759 --> 00:05:03,600 fusion of industry 4.0 and industrial

90 00:05:03,600 --> 00:05:08,880 iot with old standards and methods

91 00:05:09,680 --> 00:05:11,919 benefits

92 00:05:11,919 --> 00:05:14,479 saving cost on repair of broken

93 00:05:14,479 --> 00:05:17,680 equipment or machine shorter or no

94 00:05:17,680 --> 00:05:20,720 outage optimized use of maintenance

95 00:05:20,720 --> 00:05:23,759 staff optimized spare part and raw

96 00:05:23,759 --> 00:05:25,520 material stocking

97 00:05:25,520 --> 00:05:28,880 integration of hybrid solutions

98 00:05:28,880 --> 00:05:31,680 as predictive maintenance do not use

99 00:05:31,680 --> 00:05:34,320 hardware modifications or tempering

100 00:05:34,320 --> 00:05:37,360 which do not cause warranty issues of

101 00:05:37,360 --> 00:05:40,639 equipments or machines

102 00:05:40,639 --> 00:05:42,800 improved remaining useful life of

103 00:05:42,800 --> 00:05:45,280 machines

104 00:05:45,360 --> 00:05:48,080 no expert knowledge required by machine

105 00:05:48,080 --> 00:05:51,080 operators

106 00:05:52,400 --> 00:05:55,680 knowledge sharing

107 00:05:58,000 --> 00:06:00,800 active participation and contribution in

108 00:06:00,800 --> 00:06:06,120 sony's presence and edge impulse forums

109 00:06:07,919 --> 00:06:10,400 in-house developed documentation

110 00:06:10,400 --> 00:06:12,319 showcasing journey of product

111 00:06:12,319 --> 00:06:15,440 development which is useful in terms of

112 00:06:15,440 --> 00:06:18,240 sharing knowledge

113 00:06:18,240 --> 00:06:21,680 target industries

114 00:06:21,680 --> 00:06:24,400 hazardous industries like construction

115 00:06:24,400 --> 00:06:27,120 chemical foundries etc where

116 00:06:27,120 --> 00:06:30,720 human-to-machine interaction is direct

117 00:06:30,720 --> 00:06:32,720 data-driven proactive maintenance

118 00:06:32,720 --> 00:06:35,280 methods to analyze condition of

119 00:06:35,280 --> 00:06:37,680 equipment and raw material to help

120 00:06:37,680 --> 00:06:39,759 predict when maintenance should be

121 00:06:39,759 --> 00:06:42,319 performed

122 00:06:43,440 --> 00:06:45,280 small scale industries using

123 00:06:45,280 --> 00:06:49,800 conventional machines or equipments

124 00:06:51,840 --> 00:06:54,880 social benefit

125 00:06:55,280 --> 00:06:59,840 according to cghr 2015 report

126 00:06:59,840 --> 00:07:03,039 in india after road traffic collisions

127 00:07:03,039 --> 00:07:05,599 workplace accidents are the next major

128 00:07:05,599 --> 00:07:08,560 cause of injuries we studied the cause

129 00:07:08,560 --> 00:07:11,199 of injuries on the machines in our

130 00:07:11,199 --> 00:07:13,680 college

131 00:07:13,680 --> 00:07:17,280 in the cnc machine the door is open

132 00:07:17,280 --> 00:07:20,000 there is no visual indicator

133 00:07:20,000 --> 00:07:23,120 very limited information on screen

134 00:07:23,120 --> 00:07:26,160 no real time feedback or status of

135 00:07:26,160 --> 00:07:29,160 machine

136 00:07:32,160 --> 00:07:34,639 in this lathe machine there is no modern

137 00:07:34,639 --> 00:07:37,120 integrations and hence predictive

138 00:07:37,120 --> 00:07:39,759 maintenance can help in predicting any

139 00:07:39,759 --> 00:07:43,400 proactive situations

140 00:07:47,360 --> 00:07:49,680 in this surface grinder machine there is

141 00:07:49,680 --> 00:07:52,720 again no visual indicator not any

142 00:07:52,720 --> 00:07:54,960 digital feedback method available on the

143 00:07:54,960 --> 00:07:57,960 machine

144 00:07:59,759 --> 00:08:02,479 predictive maintenance can help lower

145 00:08:02,479 --> 00:08:06,000 down number of workplace accidents which

146 00:08:06,000 --> 00:08:09,759 lower down absenteeism medical claims

147 00:08:09,759 --> 00:08:13,080 and compensation

148 00:08:16,960 --> 00:08:18,800 product cost

149 00:08:18,800 --> 00:08:21,360 it is divided into two parts

150 00:08:21,360 --> 00:08:23,440 hardware as a service

151 00:08:23,440 --> 00:08:26,639 and subscription model

152 00:08:26,639 --> 00:08:30,560 hardware cost include module solution

153 00:08:30,560 --> 00:08:34,159 which is 10k industrial upgradation

154 00:08:34,159 --> 00:08:37,360 which include module commercialization

155 00:08:37,360 --> 00:08:40,799 certifications and ipr which will cost

156 00:08:40,799 --> 00:08:44,240 50k and hybrid compatibility which

157 00:08:44,240 --> 00:08:47,120 include plug-and-play compatibility that

158 00:08:47,120 --> 00:08:50,160 is hardware and connectivity which cost

159 00:08:50,160 --> 00:08:53,519 40k total hardware cost will be well

160 00:08:53,519 --> 00:08:56,240 lack inr per module

161 00:08:56,240 --> 00:08:58,640 next is subscription model which

162 00:08:58,640 --> 00:09:00,640 contains amc

163 00:09:00,640 --> 00:09:02,880 that offers hardware service

164 00:09:02,880 --> 00:09:05,920 compatibility validation and training

165 00:09:05,920 --> 00:09:09,360 which will be 10k monthly subscription

166 00:09:09,360 --> 00:09:12,320 which includes over-the-air updates

167 00:09:12,320 --> 00:09:15,680 servicing and validation which will cost

168 00:09:15,680 --> 00:09:19,360 5k per module warranty guarantee and

169 00:09:19,360 --> 00:09:22,640 replacement will be integrated platform

170 00:09:22,640 --> 00:09:25,600 as a service which include training and

171 00:09:25,600 --> 00:09:27,920 software customization

172 00:09:27,920 --> 00:09:32,160 thank you iit madras pravartak and sony

173 00:09:32,160 --> 00:09:34,959 india for providing this wonderful

174 00:09:34,959 --> 00:09:37,760 opportunity