*2.2. Electronic Design*

There are different approaches to measure load on a cane. The most usual ones rely on force sensors on handgrip [22,23] and/or force sensors replacing part of the shaft [27]. Some of these solutions may affect the center of gravity and/or ergonomics of the cane depending on the sensor location, as commented. Nevertheless, the main drawback of most existing approaches is that they rely on relatively expensive electronic components, such as piezoelectric quartz force link with in-line amplifier (Kistler Instrument Corp., Novi, MI, USA) [27] or array of force sensors (FSR 402, Interlink Electronics, Los Angeles, CA, USA) [22], reducing affordability for end-users. Additionally, solutions do not deal with the charging process, which can be a challenge for users with vision and/or cognitive impairments.

The main problem with force sensors is that their price increases significantly with their measurement range. Our proposed solution relies on combining two cheap force sensors at different depths into a single piece (**B** in Figure 1b) to increase their original range (see next section). We have designed a specific plastic piece to arrange them perfectly inside the shaft using a 3D printer. This approach is affordable because it only involves: two low-cost sensors (FSR 402, Interlink Electronics, Los Angeles, CA, USA), an inverter (TC7662B, Microchip Technology, Chandler, AZ, USA), an array of operational amplifiers (OPA347, Texas Instruments, Dallas, TX, USA) and some resistors and capacitors. Figure 2 shows the sensor board. The input signals to FS402 are inverted −*Vref* using TC7662B circuit. The operational amplifiers U1, U2, and U3 are adjusted with a 2.4 kΩ and 100 Ω resistors respectively to obtain the higher range of support measurement. The filtered output of FSR 402 force sensors U1, U2 are added to obtain the combination of both sensors U3. The sensor board provides a 50 Hz output.

**Figure 2.** Sensor electronic board. The input signals to FS402 sensors are inverted. *U*1, *U*2 filter the force sensors outputs. Then, *U*3 sums both signal to obtain the output.

We work with a BLE nano v2 microcontroller (nrf51822, Nordic Semiconductor, Norway). It transmits packets of 20 Bytes (8 readings) at 6.25 Hz to a paired device. Long-term monitoring requires a working plan for at least 12 h without recharging. For this reason, we have chosen a standard Qi Wireless Charger receiver with a Lipo charger (TP4056, Nanjing Top Power ASIC Corp, Nankin, China) connected to a 1S1P 500 mAh Lipo battery that reportedly provides over 4 days of use without recharging.

To heuristically calculate the system battery working life, we have performed 10 runs of the same test, consisting of feeding the cane non-stop until communication with the microcontroller is cut off, i.e., a paired device does not receive packets any longer. The battery working life is inversely proportional to the load, i.e., higher loads will lead to shorter battery life. Hence, we have used a static weight on the cane to simulate the user's load. The period of loading versus non-loading and the amount of loaded weight have been mechanically changed to simulate a large variety of users' behaviors. After several tests, the range battery working life turned out to be 82–89 h, i.e., more than 3 days of continuous monitoring.

Overall, the cost of the proposed system is cheap when compared to other solutions (less than USD 100 in total), plus it requires only minor modifications in the cane.
