Raspberry Pi (referred as RPi or RPI) and its clones i.e. Orange Pi, Banana Pi, Ordroid, Cubie, Olimex, are the class of devices located somewhere between low constraint IoT boards and regular PC/Mac machines.

Those devices technically are very close to smartphones and are far away from energy-efficient IoT solutions powered by a single battery that lasts for weeks or even years. They need DC current (usually 5-12V and about 2-3W total, with external power adapter) but still far less than even most efficient notebooks or PCs. They also use an operating system booted from storage like regular PC computers - usually from flashed MicroSD card or embedded eMMC flash. The OS is mostly Linux based but there do exist Microsoft Windows for certified Raspberry Pi devices. This is how this class of devices differ from i.e. Arduino, where software is in SoC model. The RPi and clones are holding one-board solution that includes a processor, memory, storage slot, USB and networking. Many devices also offer hardware-based graphics acceleration, usually integrated with the processor. Some devices like Orange Pi frequently offer integrated flash for OS storage, so you do not necessarily need to boot and use an external flash like USB dongle or TransFlash card.

What is pretty similar to the low-power, constrained IoT boards is RPi and clones offer GPIO and you do can connect various sensors and expansion boards (called here “hats”) and you have a wide choice of operating systems and modules. You can also extend the hardware by connecting hats that offer to sense and actuating but sometimes advanced computing like dedicated coprocessors or FPGA-based AI. Interestingly, their GPIOs usually offer (among others) popular protocols like I2C, SPI, One-Wire, so you can directly connect many sensors known as designed for Arduino-compatible development boards. This way you can use those boards like regular IoT devices with integrated networking capabilities, similar to i.e. ESP chips.

What is much different from low constrained IoT devices is that they offer at least a command terminal you can connect to, and also most boards offer a capability to connect it to the external display via HDMI, analogue output or dedicated connector for LCD display. They also offer the ability to interact with HID devices like regular keyboards, mouse, via USB but also wireless i.e. Bluetooth connection. Of course, those features are dependent on operating systems. Manufacturers usually are trying to keep those development boards as small as possible and it is a case that among high-end devices they also offer some constrained solution yet usually 50% smaller in size and power consumption (i.e. RPi zero).

Being so far from the low-power, constrained IoT devices does not exclude them from IoT devices, however. They find their application everywhere, when there is need for higher processing resources (i.e. voice recognition), high capacity and complex networking operations, i.e. gatewaying other devices to the Internet, convert networking protocols, implement software-based or hardware-assisted Artificial Intelligence, implementing rich user interface (GUI) where constrained devices are not powerful enough to fulfill the requirements yet there is still limited power source or there is not a need to set up a regular, PC-based solution, because of its cost. Most of the devices belonging to this class still can be switched to low power consumption modes, where low power means a dozen of mA here.

On the other hand, most modern representatives of those devices are powered with multicore processors and large RAM and are powerful enough to replace the desktop computer in daily operations like web browsing, multimedia playback, software development and so on.