Virtual Lineslide Inspection

A desktop based application for Railway Engineers for the UK's Largest Railway Operator to do a virtual inspection of railway tracks for vegetation encroachment.

Summary

The requirement was to envision a desktop based application for Railway Inspection Engineers to do a virtual inspection on unmanned and remote railway tracks to monitor vegetation encroachment.

Based on this observations on a specific area/side of the track, further actions needs to be recommended by the system using Machine learning algorithm.

This application utilises LiDar (Light Detection and Ranging) data and video recordings from the driver side of the train engine-room to perform the virtual inspection. This comparison helps the engineer to perform a virtual analysis of the amount of vegetations encroaching on the tracks at a specific location and helps in the creation and management of maintenance Work-orders by sitting in the comfort of their office located elsewhere.

Role

UI/UX Lead

Project Management

My Responsibilities

Discovery and requirements gathering
Wire-framing: Paper and Digital
Low-fidelity wireframes and prototypes
High-fidelity layouts
User Interface Design
Accounting for accessibility
Usability testing and validation

Team Size

3 UX (1 Lead), 2 BA, 1 SME

Timeline

(Feb to November )2022-10 Months

Target Users

Inspection Engineers

Back Office Users

Tools

Adode XD, Photoshop, Illustrator, Miro Board,

MS Teams

Problem Overview

Vegetations encroaching on to the lineside of railways has always been a threat to the visibility of Train drivers as well as the smooth running of trains on tracks. As these vegetations keeps on growing everyday, their encroachments are not monitored periodically they can interfere with the visibility of both train drivers as well cause serious blockages on train tracks. Since the railway network spans over 20,000 miles over UK, it is physically impossible to inspected every area for vegetation overgrowth or encroachments.

Currently these inspections are done manually by periodic physical inspections and the time, manpower and infrastructure required are massive and less efficient.

Once these Inspections are completed a work-order is crated by the inspector for the incident site and then the maintenance team can go and do there maintenance activities and close the work-order.

The Challenge

The client is looking for a solution to digitalise the above said activity and is looking on to the possibility of virtual lineside inspection from their office building. For this they will be utilising satellite imagery data (LiDar) of the physical site and the video recording from the camera located at the driver side of train Engines. The idea is to observe these two data source and analyse them frame by frame and record the finding on to a pre defined check list. Based on the checklist a work-order will be initiated highlighting appropriate actions by the inspector who does these virtual inspections. This work-order will then be passed on to the right maintenance team for actual maintenance work on physical site.

LiDar (Publicly Sourced)

Driver side Video (Publicly Sourced)

#1 Discovery

In this phase we get to know each other and conduct in-depth research.We identify the business problems, its objectives and what makes your brand unique. We also study the market, analyze competition and figure out users' needs and pain points.

Workshops

We started with a series of regular team discussions and workshops with SMEs and stakeholders to better understand business requirements, gather existing stakeholder knowledge, current limitations, expectations and create consensus on plans and priorities for a product, its direction and key milestones.

Needs

The inspection plan to include visual assessment of third party vegetation where it poses a risk to the railway.
Review the plan and associated frequencies of inspection annually to assess if the frequency of inspection is sufficient to control tree risk.
The inspection shall assess risk posed by vegetation within the immediate action, action and alert zones
Where unsafe situations are found during the inspection, call Control and request protection for the railway or third party.

Wants

Output of the vegetation inspection shall be recorded against every eighth of a mile for each side of the track
Digital photos should be taken to support the inspection and where work is required. This should include where it is necessary to establish the location of follow on activities.
The Inspector shall assign the appropriate response (Pre defined) for vegetation in accordance the inspection findings.
Any trees identified as being potentially hazardous, with the capability to cause derailment or harm, shall be assessed and recorded.
Record and assign a priority within its current growth season for giant hogweed that is growing in locations accessible to those on Rail land or the public.

Desires

Trees that are recognised or have the potential to be nationally important, legally protected or of local importance shall be recorded.
Where a unique identification number is required for a tree and does not exist from a previous inspection attach a tag to the tree and record the unique identification number.
Leaf fall inspections shall be carried out to assess the severity of leaf fall expected during the Autumn period on operational lines for each eighth of a mile section.
A plan shall be produced and managed so that repeat visits to the same locations are avoided. The plan shall be reviewed annually.

Setting directions for MVP

Field Visits

A field visit was conducted with SMEs and On-foot Field Engineers to understand and record various activities that are currently done and noted as part of physical inspection. This exercise gave us the opportunity to empathise with users and understand what was expected to be digitalised from a users perceptive of the application.

Observations

On-foot inspectors carries around a tablet to record their on-site inspection.
The amount of On-foot inspections are limited to few miles based on climatic conditions and other external factors.
Areas with low or no internet coverages causes difficulty in uploading the inspection summary to the central servers.
A faulty hand held device can affect/delay the defined target for inspection.

Contextual User Interviews

In the empathise stage, we also organised and conducted few structured contextual user interviews with the on-foot inspectors, allowing us to understand participants' current behaviors, expectations, or frustrations which in-turn helped us to set expectations for the new product . I chose a series of 5 questions, both qualitative and quantitative in nature.

Findings

On-foot inspectors carries around a tablet to record their on-site inspection.
The amount of On-foot inspections are limited to few miles based on climatic conditions and other external factors.
Areas with low or no internet coverages causes difficulty in uploading the inspection summary to the central servers.
A faulty hand held device can affect/delay the defined target for inspection.

#2 Strategy

This is the phase we worked on setting up correct goals, determine product features and estimate the scope of work that will provide an optimal business impact. We conducted thorough Brainstorming, user research, including developing personas, user stories, an empathy map, and a user journey map.

Designing for a data-heavy platform resembles the work of a sculptor – you start with a massive block and then slice it away until you reach just the right thing. Data-driven applications are different in a way, that less data often means a less practical and useful solution. Keeping all the data you can, however, yields an unworkable app. It’s all about finding a balance, prioritizing and organizing features according to the main use case.

Understanding the User Groups & Defining Personas

Synthesising the data from the interviews, research and identifying the pain points / opportunities we were able to form two distinct behavioural user personas and their pain points. This subsequently allowed us to write definitive problem and hypothesis statements, and formulate a user journey.These persona helps give direction to what the intention and core needs of the design solution should be.

Defining Task flow

The research collected up to this point suggests that many people are not 100% satisfied with how their newsfeed is being treated and the amount of irrelevant content that is appearing.
To summarize the findings, I created a pain points and opportunities chart to use as the foundation of ideating my solution.

User flow

We then proceed ahead to envision the user flow to describe the typical or ideal set of steps needed to accomplish a common task performed with the product by each user group.

01

02

Customer Journey Mapping

Screenshot 2023-08-02 at 10.10.32 am.png

#3 Design

Finally, we are ready to move onto design. Leveraging research data and project objectives, we build the most comforting, engaging and desirable experience for users, helping them achieve their goals as quickly as possible with minimum fuss.

Accessibility Considerations

01

We made sure all our design library colors and their combinations were AA to AAA accessible and all passed the minimum contrast by using a third part engine.

02

We made sure that our text, buttons and icons were of an appropriate size to be easily legible, no matter upon what screen size a user was viewing and using the site.

03

We also explored annotating our prototypes for web accessibility, based on WCAG guidelines. These annotations show the engineer the linear focus order, or traversal order that assistive technology (such as a screen-reader), would move through the form.

Branding Guidlines

As the streamlined and integrated flow became more and more clear, I began sketching wireframes of the new solution. The purpose of the low fidelity sketches was mainly to ensure hierarchy, design goals, and make sure that I was using the proper components for each function.

Exploration

As the streamlined and integrated flow became more and more clear, we began sketching wireframe layouts of the new solution. The idea was to accommodate large data sets in the best possible layout for users to perform their activity efficiently. First we defined possible layouts and did an analysis to find which layout wound best suit our requirement.

Layout 1

Layout 2

Layout 3

Layout 4

Anatomy of UI

I started working on a few key screens in low-fidelity to find the best solution in terms of usability and scalability. After testing and refining, the overall layout started to shape into the unified structure that can be applied for different types of users and their use cases.

#4 Build

Once we received wireframe sign-off from stakeholders the design where handed over to the development team building the MVP. The development was divided into three releases based on user story prioritisation. After each release the MVP was tested and the feedback was used again to enhance the next release build.

Final Outcome

There is a strong trend of reducing complexities in design. This works in consumer apps with single and simpler goals, but when you design for enterprise products, not always you can (and should) do it. We built a product for a specific audience that values functionality over the look and feel, has certain expectations and goals.

Managing a small agile team allowed me to move faster without noticeable compromises either on the design or tech side. We managed to introduce a significant amount of improvements compared to what the industry operates with today. At times, there were hundreds of challenges building certain elements that fall under industry regulations in such a short time.

#5 Launch & Learn

The design is a constantly evolving and improving process thus cannot be 'completely done'. Once the product is released and users start using it, we get a great opportunity to review its performance, aggregate feedback and improve based on user needs and business goals. We followed the same process to evolve the product even better.

Usability Testing

In the testing stage I focused on the Usability Study parameters, and followed with an in-depth research study plan. This comprehensive stage also included synthesizing the found data with a note taking spreadsheet, a pattern identification spreadsheet, an insight identification template, an affinity map, and finally my usability study findings.

Key Insights

Users said they were very overwhelmed with the amount of content on each page. Too much type causes user to quickly lose interest. Information overkill.
We observed some users getting stuck in the Inspection Process, slowed down by too many questions about preferences and the tedium of checkboxes.
Side by side comparison of FFV and LiDAR data really helped users in performing an efficient virtual Inspection.

Short-term Goals

Continue to user test the new design and iterate new versions of the design as needed for further testing.
Work with Dev team to implement design changes and determine whether some of the new features or functions can be done with the existing layout.
Iterate a tablet version of the responsive site.

Long-term Goals

Make some functional changes to the app to improve user experience.
Include provisions to accommodate WAIF, MST already applied by other departments on the existing section of tracks in focus.

Supporting user stories in Azure DevOps by providing relevant wireframes, acceptance criteria and prototype link references.
Make custom wireframes on the fly for special scenarios to speed by engineering process with out delay.
Participate in build meeting to supervise the build progress
Test build to validate the user story alignment with the UX guidelines and interactions.

#6 UI Dev Support

After the UX delivery was completed and handed over to the development teams, major follow up was needed to ensure all the conditions written in acceptance criteria were followed in terms of Design guidelines, Accessibility and Interactions. A good deal of time was spent working on a design integration strategy with the engineering team. We devised a phased approach that would have minimal impact on the work in progress, and allow everyone to become familiarised with the fabric of design system. Some of the main production support activities are-

VLI was one of the biggest and complex project which I did recently. Here we had the opportunity to conduct detailed primary research before jumping into a solution. We also had the opportunity to test with real potential users. I think the research, interviews and testing were really helpful in putting away all the assumptions and focusing on the core needs of the target audience. It also helped us better understand the constraints of the project.

Collaboration is key

We each took responsibility for designing different user group but always reviewed our progress twice a week with each other to make sure the product is cohesive throughout.

Viewing problems from a user’s perspective

Some design decisions may seem too obvious as a designer, but may not be very obvious to a user. User testing really helped us in making informed design decisions.

Learned new process for building product

Learned properly planning a sitemap and using System Requirement Chart was really helpful in prioritizing important features and scoping the product for a simple MVP.

Reflection

Tinton was the Lead UX for this project from the start. His UX expertise also hinted on good communication skills and attention to detail. Both of which turned out to be very helpful during the project progress. He was very committed to getting the design perfect in every detail. He has a good sense for layout and white space. He went above and beyond with adopting workflow processes. He coordinated and managed workshops and made sure it was always engaging. Tinton also took feedback from stakeholders on board and has always demonstrated a quick turnaround responding to feedbacks. Business peers gave his design prototype great reviews.."

Client Feedback

Recognition

This product was one of the 5 products under "Insight Program" which bagged the Digital Technology Leader Award for the Best Machine Learning Projects 2023. A Cognizant and Network Rail collaboration.

Thank you for your time.