CONFERENCE DAYS 2024: Data is Oxygen – AI-based Analysis of Vehicle Data

At this year's CONFERENCE DAYs, which HUSS-VERLAG offers every year as a digital B2B knowledge and networking event for the logistics & intralogistics, truck & bus as well as automotive & taxi sectors, a session was also devoted to the latest data analysis methods for fleet efficiency improvement. Michael Schmidmeier, Business Development Manager at the Canadian specialist for telematics solutions in the transport sector Geotab, was invited.

The company has developed an open fleet management platform that aims to optimize fleets using a variety of customizable and state-of-the-art analysis methods. According to their own statements, the solution is already used by more than 50,000 customers worldwide. In his presentation, Michael Schmidmeier demonstrated the advantages and functionality of the system and explained the contribution and benefits of Artificial Intelligence (AI) to its operation.

Renowned Customer Base

As the self-proclaimed global industry leader in connected transport and vehicle tracking solutions, the company founded in 2000 supports transport businesses with data-based information to streamline their operations. Among the clients of the company, which is highly valued in the North American market according to its own statement, are globally active corporations such as logistics provider DHL, car rental company Europcar, as well as Deutsche Bahn and Swiss Railways.

One reason for the success of the applications is likely due to the provider owning all intellectual property rights to device hardware and firmware, server software, and client-side tools, allowing it to respond particularly quickly to customer requirements and market opportunities and to expand its functionalities, as stated by the company headquartered in the eastern Canadian province of Ontario. This unique feature gives it an advantage over providers who outsource intellectual property and development.

AI-Enabled Platform for Connected and Sustainable Fleets

Geotab integrates all pertinent data from the devices important to the respective company, "regardless of the source." The transmission of "high-quality, precise" information from GPS devices, for example, is ensured by a curve-based algorithm, which is intended to enable intelligent decision-making on an AI-based platform. The application also utilizes previously integrated data from many original equipment manufacturers (OEMs) as a working basis.

According to Schmidmeier, all these data are then categorized based on the six "pillars" of the Geotab application, focusing on their use for productivity, safety, sustainability, or scalability, among other aspects. This allows for the analysis of operational procedures, diagnostics, and the initiation of optimizations. One advantage of the open interface is that the information is accessible to customers upon request.

Multilayered Processes

For example, the goal "reduce emissions" can be selected from the various functions. In this case, a multilayered process is followed, which aims to filter out the best solution based on the wealth of collected data. Steps such as "Evaluate – Implement – Promote – Fulfill" are applied in this area. For an even more robust assessment, data from comparable fleets are also incorporated into the evaluation.

Optimization Points

As a result of the individually created evaluation, it may reveal, for example, that one's own fleet would operate more economically with fewer vehicles, thus recommending a reduction of the fleet. Other adjustment points for optimization, due to the extensive data covering nearly all aspects of operation, include energy-efficient route planning or improving the driver profile. When reading data from electric vehicles, factors such as power consumption and battery condition over the entire usage period are naturally also recorded.

Bagel Analogy

According to Schmidmeier, the building blocks of the system can be compared to a bread recipe. For example, a bagel is fabricated in certain steps with corresponding ingredients using specific tools, whereby no single element can be missing, and the quality is crucial to achieve a good result. Similarly, optimizing the vehicle fleet requires certain indispensable high-quality foundations and procedures to achieve an optimal product, or in Schmidmeier's words, "build the bridge to telematics" to evaluate the data.

For data analysis, high-quality, precise GPS data and specific vehicle data for the problem to be investigated are needed as ingredients. The recipe means integrating the data depending on which factor is currently being considered. The right tool means the platform where data analysis takes place, such as the cloud in this case. However, the right location must also be included in the evaluation, meaning that the currently applicable legal requirements of the respective region must be taken into account.

Instant Fleet Scanning with AI

These classifications serve as a kind of framework for the data pool, which is then used as "feed" for the AI system. If the relevant information is available for all required aspects, various insights can be gained, providing fleet operators with a detailed overview of their operations.

For example, evaluation methods can also be created in this way. One possibility mentioned by the expert is reward systems for correct driving behavior. Schmidmeier:

"If all data is available, a powerful AI-based system can emerge from it."

The application's dashboard clearly presents the results with an integrated camera video system. The AI-supported video telematics provides an overall view of the fleet, meaning it registers whether a driver is smoking, using their phone, is inattentive or otherwise distracted – all these, along with vehicle data such as transmission oil level, are displayed on the screen.

Driver Coaching

The data-based fleet inspection hardly misses any flaw, so that errors or damages can be detected and rectified beforehand, preventing vehicle breakdown. The AI-supported system structure also allows for early prediction of possible collisions and other events related to vehicle operation. Users can also investigate specific aspects individually through the platform and configure the application according to certain rules. The process is then controlled by AI, and upon activation, all necessary data is automatically recorded. This way, collisions can be avoided.

Actions during driving, such as using a smartphone, are precisely recorded and noted in the driving report. Thus, the recording can also be used to optimize driving behavior.

AI and ChatGPT

Schmidmeier referred to a special application, "Geotab Ace," the "first AI assistant related to vehicle data," or an AI-supported, dialog-capable query system.

The particular benefit is that compiling all the necessary data individually would be very time-consuming; however, using AI and the AI-based speech and text system ChatGPT-4, natural language can be used to interact with telematics data to gain insights into the fleet's condition. Similar to well-known voice assistants like Alexa or Siri, it enables communication with the system by voice and asking specific questions, such as the fleet's fuel consumption in 2023, or which vehicles were speeding on a particular day. To provide answers, the system draws from the extensively fed data pool. The system should also respond to follow-up questions to clarify facts. To verify the accuracy of the answers, the SQL (Structured Query Language) database query system can be utilized.

Further detailed queries, such as which vehicle has the lowest battery charge level, and many others, are possible in this way. As all conceivable useful questions have been embedded in the AI function—Schmidmeier speaks of about 100 currently—various context-related insights are now possible. Moreover, the application can deliver near real-time and hassle-free information on almost any fleet area, generating corresponding clear reports upon request. What used to take hours of analysis is now completed in seconds. The "generative AI" reduces the steps and provides simplified data analysis.

Data Protection

At the end of the session, questions primarily about data protection were addressed. According to Schmidmeier, all data in Geotab Ace remains confidential and can only be queried through this platform. Individual settings, such as regarding privacy in the cabin camera evaluation, allow the camera function to be turned off. The video camera records only "event-based," with a ring buffer where the information is available only for a limited period.

Geotab Ace works with OpenAI's "Large Language Model," meaning that customer query data stays within a closed circle and is not shared externally. The system also does not interfere with the user's decision-making: "the system does not give advice," it says. According to Schmidmeier, all results remain the customer's property and are for their information only. The video function can be configured heavily to meet individual needs. For greater acceptance of the application, a rating system could be set up, benefiting the drivers.