**1. Introduction**

A 2017 United Nations report on world population ageing [1] shows that the number of persons over 60 years in 2050 will double with respect to 2017, i.e., in 2050 one out of five people worldwide will be seniors. Indeed, population ageing is already significant in Europe and Northern America, where currently more than one out of five persons are already over 60. Hence, healthy ageing has become the main concern. One of the main tools to promote healthy ageing is, reportedly, monitoring and feedback to users [2], especially for the most vulnerable population, like persons with some degree of disability. Mobility monitoring has attracted major interest, as it is fundamental to keep a healthy and active lifestyle and to remain autonomous. The simpler, most popular approach to

mobility monitoring are activity/fitness trackers, that provide data on distance walked or run, calorie consumption, and, in some cases heartbeat or quality of sleep. However, parameters like distance walked are only indirectly related to a person's condition. Alternatively, gait (and posture) analysis is widely acknowledged as a clinically useful tool for identifying problems with mobility. Poor gait may lead to musculoskeletal pain in different parts of the body, whereas a good gait will reduce the risk of joint problems, help good recovery after injury and/or surgery, improve mobility in the elderly and reduce fall risk [3]. Consequently, gait monitoring has attracted considerable attention.

Clinically, gait analysis is performed via scales like the Tinetti Mobility Test [4]: experts observe patients performing different tasks and manually provide scores for several indicators. However, this process is slow and requires significant time from patient and clinician. There are technology-based alternatives to this approach, like using force/pressure sensors on a walking surface like a treadmill or a walkway [5,6] or optical motion capture systems [7–12]. However, people need to be assessed in specific installations, and experts are often needed to set and/or calibrate equipment. Gait assessment during Activities of Daily Living (ADLs) may be achieved using wearable sensors, e.g., sole pressure sensors [13,14] and/or inertial sensors [15–17]. However, some of these wearables may not be comfortable nor easy to attach/calibrate [18] so people often do not carry them. Sensors can instead be attached to mobility aids if people require them for their ADLs [19,20]. Indeed, smart wheelchairs and rollators often include different sensors. However, attaching them to canes, the most spread mobility aid, with minor alteration of their ergonomics (Modifications that affect ergonomics or center of gravity in walking aids require full analysis, validation, and standardization of the modified device, a process usually costly and slow) is usually harder.

Load on a cane can be used to estimate a person's partial weight-bearing by calculating the differences between the users' weight and the load. Partial weight-bearing provides information about how much load the user supports on their affected side. This is an indirect measure of their condition and it can be used to monitor their evolution, provide feedback and, in clinical treatments, to adapt their rehabilitation process [21]. There are different solutions to measure load on a cane. In specific, fixed environments, a walkway with a dense matrix of force sensors can be used, e.g., StridewayTM (Strideway System; Tekscan, Inc., South Boston, MA, USA) (Strideway System. Available online at: https://www.tekscan.com/products-solutions/systems/strideway-system?tab=applications). These systems measure weight-bearing accurately and, hence, the cane load by default. However, they only measure a very limited number of meters on a straight line, plus they are expensive and, as aforementioned, constrained to specific environments. Hence, this work focuses uniquely on onboard cane sensors. Currently, onboard sensors on most smart canes are located either on the handgrip [22,23], shaft [23–27] or tip [22,24]. Every location has advantages and disadvantages. Placing sensors on the handgrip or tip may involve major cane modifications [22,23]. As both locations affect how users support their weight, these modifications must be ergonomic. Unfortunately, granting ergonomics requires an extensively validation process. On the other hand, the shaft allows more space to place the electronic. However, this approach may involve changes in the cane center of gravity and, in its weight [23–25,27].

This work proposes an affordable add-on module for long-term monitoring of load on cane. The obtained value can be used for long-term assessment of users' condition, either for preventive healthcare or for evaluation of degenerative or rehabilitation processes. The module has been designed to be extensively available to as many potential users as possible, so its main goals are: (i) compatibility with existing commercial canes; (ii) to avoid any impact on cane ergonomics; (iii) to allow continuous, long-term use; and (iv) to keep the global cost as low as possible. Additionally, we plan to release the proposed system under a Creative Commons License to boost its reach and impact. Section 2 describes the proposed system, i.e., mechanical design and electronics. The load cane estimation is described in Section 3. Section 4 describes our methodology. Tests to validate that the module provides meaningful load estimation are presented in Section 5. Section 6 discusses our results. Finally, Section 7 presents the conclusions and future work.
