In the quest for perfectly baked cookies, relying solely on cooking time or color can be surprisingly unreliable. A recent study from Indonesia's Institut Teknologi Sepuluh Nopember (ITS) explores an innovative solution—an electronic nose that can identify cookie doneness by aroma, offering a more consistent and automated way to assess food quality during baking.

The Challenge of Measuring Doneness

Baking cookies is both an art and a science. Traditionally, bakers use time, temperature and visual cues like color or texture to judge when cookies are ready. However, many cookies look deceptively similar at different stages of baking. This can lead to inconsistent quality, especially in large-scale production settings. Furthermore, techniques like spectroscopy or gas chromatography, though accurate, are costly, time-consuming and require trained personnel.

Enter the Electronic Nose

An electronic nose (e-nose) mimics the human olfactory system using gas sensors and artificial intelligence (AI) to detect odors. Unlike traditional gas analysis tools that identify specific chemical compounds, an e-nose reads the general aroma pattern of the air around it. This technology has already proven useful in industries such as tea processing, coffee roasting and even medical diagnostics.

In this study, researchers developed an e-nose system specifically for detecting the cooking stages of cookies. Their approach involved a sensor array, an Arduino microcontroller and AI models to classify the aroma signals into undercooked, cooked, or overcooked stages.

Method and Innovation

The research team tested 20 uniformly prepared chocolate chip cookies, baking each for up to 45 minutes at 150°C. The e-nose setup included 20 gas sensors designed to detect a variety of volatile organic compounds released during baking. These sensors were paired with several machine learning algorithms, including convolutional neural networks (CNNs), support vector machines (SVMs) and others.

To optimize the system, the researchers applied statistical techniques like correlation analysis and distinguishing rate (DR) calculations. These helped reduce the number of sensors while maintaining accuracy. After testing, only six sensors were deemed necessary to achieve high performance, leading to a more compact and cost-effective device.

AI-Powered Accuracy

Among the 10 AI models tested, CNNs demonstrated the highest performance in classifying cookie doneness based on aroma patterns. The final system using six gas sensors and CNN achieved:

• Accuracy: 90.0%

• Precision: 89.7%

• Recall: 92.6%

• F1-score: 90.2%

  
  

This shows the system’s strong ability to detect when cookies are just right, helping ensure consistent product quality.

Practical Application and Future Potential

To validate the system, the team tested it on 20 new cookie samples baked at random times and temperatures. The e-nose successfully identified their doneness levels. It even adjusted to different oven temperatures (150°C, 175°C, 200°C), signaling when cookies reached optimal doneness with minimal deviation in quality, as confirmed by taste testers and objective color/moisture data.

This technology could be integrated into electric ovens to automate baking quality, reducing reliance on human judgment. Such integration would be especially useful in commercial kitchens, food factories, or even smart homes.

In short the development of an electronic nose capable of accurately detecting cookie doneness is a leap forward for automated food quality control. This system not only enhances consistency in baked goods but also offers a low-cost, scalable solution for modern kitchens.

Future research will likely focus on miniaturizing the system and improving real-time responsiveness, potentially revolutionizing how we bake.

References Rivai, M., & Aulia, D. (2024). Use of Electronic Nose to Identify Levels of Cooking Cookies. IEEE Access, 12, 97235–97247. https://doi.org/10.1109/ACCESS.2024.3428322