1 of 36

Objective

The primary purpose of the SDR presentation is for you to show your ability to communicate your work to others effectively in a conference-style presentation form. Its secondary purpose is to give you an opportunity to present your project implementation progress and issues to the class and instructors.

Presentation

Each team will have 27 minutes total, 17 minutes for presentation followed by 10 minutes of Q & A. Use, but do not exceed by more than two minutes, your full 17 minutes of presentation time - there is plenty to describe. Plan on using a team laptop to present, since ensuring that videos, or animations, or other content elements run on another laptop can be tricky.

2 of 36

General

1. A reasonable rule of thumb for presentations is one slide per minute, but this will depend on various factors, such as whether you have animations, so be guided by the time limit, and create however many slides fit into that.

2. Everyone on the team plays a role in the presentation.

3. Attend all presentations so all teams have an audience. We will start on time, so be there a few minutes early. It goes without saying that you must be there for your team's presentation. If you are absent on the day you are not presenting, you will lose 2 out of the 20 possible points.

4. Do not use physical props to describe concepts. The idea is to give a talk as you would at a technical conference, or within a company to your CEO or CTO; in both cases, brevity and focus are crucial.

5. Use videos, animations, and any other illustrative presentation techniques that Powerpoint or other presentation software allows.

6. Assume a general technical audience; i.e., don't assume your listeners have special knowledge.

7. In general, and especially for elements that would be too lengthy if presented in full (e.g. system requirements or multi-level functional architecture), pretend you are presenting to your company's CEO to get permission to build the system - what do you think he or she would like to hear most?

3 of 36

System Development Review

Corinne Alini, Raj Basu, Grace Kimes, Jacob Miller, Jonathan Schwartz

4 of 36

Project Description

User Needs:

Safe disinfection
Reliable disinfection
Cost-effective disinfection
Traceable disinfection

Our Solution: Mobile robot equipped with electrostatic sprayer

Safe to use around people
CDC approved disinfection technique
Cost-effective robot design
Logs disinfection data

5 of 36

Introducing

6 of 36

Use Case

Detect hotspots
Estimate 3D position of hotspots
Navigate to hotspots
Spray hotspots with disinfection solution
Log disinfection data

7 of 36

Performance Requirements

The system will:

Localize within environment within 10 cm accuracy
Detect hotspots with 50% mAP
Estimate position of hotspots within +/- 10 cm accuracy when within 1 meter of the hotspot
Maintain knowledge of hotspot positions as robot navigates at a frequency of 10 Hz
Navigate to within 0.25 m and 15° of hotspots on hard, flat indoor floors reliably at 0.25 m/s
Move spray nozzle with 1 degree of freedom
Spray hotspots with 0.5 m² coverage
Log disinfection data while operating

Hotspot

Door handles and light switches

8 of 36

Non-Functional Requirements

The system will:

Be low-cost (raw materials < $10,000)
Have 1 hour of runtime assuming full payload
Minimally-harmful disinfection technique

9 of 36

Hotspot Detection

Responsible for:

Detecting 2D bounding boxes for light switches and door handles

Current status: complete
Challenges faced: collecting data set
Remaining challenges: none

10 of 36

mAP This Version:

Overall: 70.8%
Tables: 50.52%
Door handles: 74.28%
Light switches: 87.08%

11 of 36

Hotspot Pose Estimation

Responsible for:

Estimating the 3D pose of hotspots in the world frame
Tracking the hotspot poses between frames, even when they are no longer in FOV

Current status: Integration testing
Challenges faced: Surface normal estimation
Remaining challenges: Testing integration with the full system

12 of 36

13 of 36

Indoor Navigation

Responsible for:

Exploring the environment
Avoiding obstacles
Navigating to hotspots

Current status:

Integration testing
Developing back up localization strategy

Challenges faced:

Unreliable hardware limiting testing

Remaining challenges:

Testing integration with the full system

14 of 36

15 of 36

Hotspot Disinfection

Responsible for:

Moving electrostatic sprayer with 1 DoF
Spraying hotspots with disinfectant solution

Current status: Complete
Challenges faced:

Noise in signals with long wires
Measure twice, cut once

Remaining challenges:

Testing integration with the full system

16 of 36

17 of 36

Disinfection Logging

Responsible for:

Logging the position and an image of each disinfected hotspot
Visually displaying logged information

Current status: Complete
Challenges faced:

Trying to develop on Windows

Remaining challenges:

Testing integration with the full system

18 of 36

19 of 36

Disinfection Logging

20 of 36

Watchdog

Responsible for:

Monitoring all critical ROS topics
Checking temperature of Jetson
Checking for stale LiDAR values (a common issue we have seen)

Current status: Done
Challenges faced: Integration with full system before full system is complete
Remaining challenges: Adding additional features to handle more failure cases

21 of 36

Schedule

22 of 36

PR #10

23 of 36

PR #11

24 of 36

Test Plan

25 of 36

Test Plan

26 of 36

Test Plan

27 of 36

Budget

Total budget: $10,500
Big-ticket purchases:

Marvelmind v100 Robot - $7,690
Marvelmind spare parts - $851
Manipulator parts - $501
Electrostatic sprayer - $429

Remaining budget: $444

28 of 36

Risk Table

Risk #	Risk	Category	Mitigation	Likelihood	Consequence	Indicator
R2	Marvelmind IPS Does Not Perform As Expected	Technical / Schedule	- Alternative SLAM algorithm using depth camera - Write dead reckoning system to act as backup	2	4	Med
R4	Water/Cleaning Fluid Gets Onto Electronics	Technical / Cost / Programmatic	- Liquid Testing Protocol - Protect Electronics from Water - Pre-test & Post-test Checklist	4	5	High
R5	Unreliable Marvelmind Hardware	Technical / Schedule / Cost	- Spares ordered - Wiring checklists	5	5	High
R8	Shortened timeline	Schedule	-Removing tables as a feature	4	4	High
R9	LIDAR Range	Technical / Programmatic	-Test Nav as soon as possible. -Research backup lidars. -use Realsense for obstacle detection	3	4	Med

29 of 36

Risk Table (Closed)

Risk #	Risk	Category	Mitigation	Likelihood	Consequence	Indicator
R1	Unable to Acquire Marvelmind Robot on Time	Technical / Schedule / Cost	- Pivot to backup platform	3	5	High
R3a	Loss of Campus Access	Technical / Schedule / Cost	-Develop out of Beeler House Lab	4	2	Med
R3b	Loss of Classroom Access	Technical / Schedule / Cost	- Budget allocated for mock demo space	5	2	Med
R6	Unable to Find Good Detection Data Set	Technical / Schedule	- Create Custom Dataset	2	3	Med
R7	Marvelmind Odometry Board Does Not Work	Technical / Schedule / Cost	- Create Custom API and with installed sensors	3	5	High

30 of 36

Risk Chart

Risk #	Risk
R2	Marvelmind IPS Does Not Perform As Expected
R4	Water/Cleaning Fluid Gets Onto Electronics
R5	Unreliable Marvelmind Hardware
R8	Shortened timeline
R9	LIDAR Range

5					R5
4			R8	R8
3				R2	R4
2
1
	1	2	3	4	5
	Consequence

Likelihood

31 of 36

Questions?

32 of 36

Appendix Slides

33 of 36

Functional Architecture

34 of 36

Cyberphysical Architecture

35 of 36

Cyberphysical Architecture

36 of 36

Cyberphysical Architecture