The Next Step for Edge AI: Giving Connected Devices Physical Awareness

Artificial intelligence at the edge is becoming increasingly capable. Cameras can identify objects and people, processors can run increasingly sophisticated AI models locally, and technologies such as 4G LTE, 5G, Wi-Fi and BLE allow connected devices to communicate almost anywhere.

But AI is only as useful as the information entering the system.

A camera gives a device vision. A microphone provides sound. Environmental sensors measure temperature, air quality or vibration. Yet many connected products still need a simple and reliable answer to a fundamental question:

Is a person actually there, where are they, and are they moving?

This is where human presence sensing can add another layer of physical intelligence to Edge AI.

AI Makes Sensors More Valuable

The development of Edge AI is changing the role of sensors.

Traditionally, a sensor might simply provide a trigger or raw measurement. Today, sensors can become part of a much larger intelligent system, working alongside cameras, microcontrollers, AI processors and connectivity technologies.

Instead of asking one technology to solve every problem, different sensing technologies can contribute the information they are best suited to provide.

A typical Edge AI system might combine:

• A camera for visual information

• Radar for human presence and movement

• Environmental sensors for local conditions

• An MCU or AI processor for decision-making

• 4G LTE, 5G, Wi-Fi or BLE for connectivity

• Cloud platforms for long-term analysis and remote management

The value comes from combining these different sources of information.

Not Every Sensor Needs to Be an AI Computer

Edge AI does not necessarily mean putting a large AI processor into every sensor.

A specialised sensor can perform one physical sensing task efficiently and provide structured information directly to a host processor.

This is where Keywave KW307 fits.

KW307 is a compact 24 GHz human presence sensor designed to provide real-time information such as:

• Human presence

• Distance

• Angle

• Movement

• Adjustable sensing range and sensitivity

Rather than generating another large stream of raw data, KW307 can provide actionable physical information to the host system through a simple UART interface.

This allows the main processor to concentrate on higher-level functions such as AI inference, vision processing, communications or system control.

Adding Physical Awareness to Edge AI

KW307 is designed to complement a wide range of edge-computing and connected-device architectures.

It can provide real-time human presence, distance, angle and movement information to host platforms built around technologies such as Qualcomm processors, NVIDIA Jetson, Raspberry Pi, STM32, Nordic and NXP MCUs, as well as systems using cellular connectivity, Wi-Fi, BLE, cameras and other sensors.

Rather than competing with these processing and connectivity technologies, KW307 adds an additional layer of physical awareness that can help the host system decide when and how higher-level AI, vision, communication or control functions should be activated.

In simple terms:

The Edge AI processor is the brain. Different sensors provide the senses. KW307 adds physical human awareness.

KW307, Cameras and Edge AI

Cameras are one of the most important sensing technologies used with Edge AI, but continuously processing video can require significant processing power, energy and data bandwidth.

Human presence sensing can provide an additional layer of intelligence around the camera.

For example, KW307 can determine that a person is present before a higher-level vision system needs to become fully active.

A system could operate in the following sequence:

KW307 detects human presence

↓

Distance, angle and movement information is sent to the host

↓

The edge processor decides whether further processing is required

↓

Camera or AI vision processing is activated when useful

↓

Important information can be transmitted through LTE, 5G, Wi-Fi or another network

This architecture can be particularly valuable in applications where power, computing resources or communication bandwidth are limited.

KW307 + 4G LTE for Remote Edge Intelligence

Cellular connectivity creates another important opportunity.

Many Edge AI systems operate outside traditional buildings or networks, including:

• Remote equipment

• Utility infrastructure

• Telecommunications sites

• Security systems

• Temporary installations

• Industrial equipment

• Agricultural environments

These systems may rely on 4G LTE or 5G rather than permanent wired connectivity.

Consider a remote monitoring system containing a camera, an Edge AI processor, a cellular modem and KW307.

Instead of continuously transmitting video or operating high-level AI processing at maximum power, KW307 can first provide information about whether meaningful human activity is occurring.

The host system could then decide to:

• Wake a camera

• Start an AI vision model

• Increase recording quality

• Activate lighting

• Trigger an alarm

• Record a local event

• Send an alert through LTE

• Upload selected data to the cloud

The objective is not simply to detect movement.

It is to give the complete system better information about when something meaningful is happening.

Sensor Fusion Creates Better Decisions

The real opportunity becomes even greater when multiple sensors work together.

Imagine an Edge AI device receiving:

KW307
Human presence, movement, distance and angle

Camera
Visual objects and activity

Environmental sensors
Temperature, air quality, humidity or vibration

Edge AI processor
Sensor fusion and local decision-making

LTE / 5G / Wi-Fi
Remote communication and management

Each technology provides a different part of the picture.

The AI system can then make decisions using several independent sources of information rather than relying on one sensor alone.

This can improve system reliability and also enable completely new product features.

Applications Beyond Security

Although camera-based security is an obvious example, the same architecture can be applied across many industries.

Smart Buildings

KW307 can provide occupancy and human-presence information to Edge AI systems controlling lighting, HVAC and building automation.

The system can understand whether spaces are genuinely occupied and adjust energy use or building functions accordingly.

Smart Lighting

Lighting systems can combine human presence information with environmental conditions, scheduling and network information.

Instead of simply switching a light on when movement occurs, the system can make more intelligent decisions based on presence, position and activity.

Remote Infrastructure

Telecommunications, energy and industrial equipment can monitor human activity around remote installations.

Important events can be processed locally and transmitted through cellular networks only when necessary.

Intelligent Cameras

Radar can provide a second sensing layer alongside video.

The camera does not necessarily need to analyse everything continuously. Human presence information can help determine when higher-level vision processing adds value.

Industrial and IoT Systems

Factories, equipment and connected appliances can use physical human-awareness information together with local AI to provide safer and more responsive interaction.

Designed for Integration

For product designers, sensing technology also needs to be practical.

KW307 is designed to simplify integration into embedded products through:

• Compact sensor design

• Low-power operation

• Real-time UART output

• SDK and API support

• Adjustable range and sensitivity

• Detection through most non-metallic materials

Keywave has also developed algorithms to help suppress common false-trigger conditions and differentiate meaningful human activity from environmental interference.

This becomes especially important when sensors form part of automated Edge AI systems where false events can create unnecessary processing, communication or power consumption.

From Connected Devices to Context-Aware Devices

The next generation of connected products will not simply collect more data.

They will increasingly understand what is happening around them and decide locally what information matters.

AI processors provide increasingly powerful intelligence.

LTE, 5G, Wi-Fi and BLE provide connectivity.

Cameras provide vision.

Other sensors provide environmental information.

Human presence sensing adds another important layer: physical awareness of people in the real world.

By providing real-time presence, distance, angle and movement information to the host system, KW307 can become part of this wider Edge AI architecture.

The objective is not to replace cameras, processors or connectivity technologies.

It is to make them more useful.

Give Edge AI better information about the physical world, and the entire system can make better decisions.

Extending reading 1: https://www.fiercesensors.com/sensors-fusion/sensors-are-more-valuable-ai-and-qualcomm-knows-it

Extending reading2: https://www.linkedin.com/posts/sponsored-edgeai-radarsoc-share-7468374888509128706-g22y/