INTRODUCTION

Two-wheeled electric auto-balancing seated vehicles (EASV) are a relatively new type of mobility device, which have the potential to improve mobility with their small footprint and zero turning radius. They may enable access to more difficult physical environments like narrow aisles, bathrooms and uneven terrain that often present challenges for power wheelchairs (PWCs) [1]. They also provide the propulsion needed to help alleviate upper extremity pain and injury found in ~68% of manual wheelchair (MWC) users [2]. Example EASV are the Nino® product [3] the Genny [4] and the Gyrolift [5], currently being developed.

Based on Segway-style technology, these devices rely on two parallel wheels, sophisticated control systems, gyroscopes and high-powered electric motors that enable balance while the user is seated (Figure 1 ). This differs greatly from conventional wheelchairs because it removes the need for casters (the main purpose of casters is to provide stability, however they can sometimes impact performance) [6]. This, and the added benefit of having the center of gravity (CoG) over the drive wheels for maximum traction make EASV an exciting step forward in mobility technology. For these same reasons, self-balancing two-wheeled devices are inherently unstable, which potentially causes a new set of risks that are poorly understood, although many mishaps are documented on YouTube. This preliminary study investigated some situations one may encounter during Nino® use that are potentially dangerous to seated users. This work will help the community understand this new technology, prioritize further research and testing goals, guide the need for user training, and inform product developers of improvements that will make EASVs more appropriate for a wider range of users.

NINO® DESIGN

The Nino® EASV is a collaboration between Nino Robotics© who designed the custom chassis and seating and Ninebot® (makers of Segway®) who provided the wheeled base (Figure 1). It was designed specifically for people with reduced mobility but who still have good seated-trunk stability and upper-limb function. A tiller-based operating method is used to turn left and right, and the system requires the user to lean forwards and back to move and brake, exactly like a Segway®. The Nino® weighs 48.5 kg, with a reported top speed of 22 km/h and range of 20-35 km. The width of the Nino® is 586 mm and depth of 740 mm, making it more similar to the projection surface of a human walking than either the MWC or PWC.

One design feature to note is the electric kickstands. These are two outriggers that extend front and back on the right side of the device. A user operated button raises them for active use and lowers them when parking the device. This allows the device to be stable when it is powered off. The outriggers extend a fixed distance, almost de-weighting the right wheel, creating a tripod with the left wheel. The kickstands are 46 cm apart.

EVALUATION

All testing was performed at the Rehabilitation Engineering Design Lab (REDLab) at the British Columbia Institute of Technology. This occurred after a user evaluation was performed by occupational therapy researchers that studied MWC users’ ability to complete a variety of wheelchair skills using the Nino® [7]. These results informed the need to perform this testing, specifically about braking and the use of the kickstands.

Braking

During the user evaluation, one major concern of participants was the stopping distance [7]. This concern was also noticed by researchers who have extensively used the device. Braking requires the user to lean backwards so that the whole device tilts. The further the lean the quicker the Nino® slows. Relative to joystick controlled devices this motion takes significant input from the user, hence the need for trunk stability. The reported emergency braking distance of the Nino® is 4 m when moving rapidly [8].

Table 1 . Median emergency flat braking distance and deceleration from varying speeds

Speed (km/h)	Median and Range Braking Distance (m)	Median and Range Deceleration (g)
20	6.1 (5 - 6.5)	0.26 (0.24 – 0.31)
15	4.8 (4.8 - 5.0)	0.19 (0.18 – 0.19)
10	2.8 (2.5 – 3.0)	0.14 (0.13 – 0.16)
5	1.5 (1.3 – 1.5)	0.07 (0.07 – 0.08)

The brake test was done with a 73 kg user.  The straight-line flat braking distance of the Nino® was tested at four speeds, 20, 15, 10 and 5 km/h (Figure 2 ). To maximize timing accuracy, each test was filmed and then replayed frame by frame. Start time was determined by commencement of user initiated backwards braking movement; end time was determined by the cessation of any forward motion. Results are shown in Table 1 . Tests were conducted five times at each speed on parking lot asphalt.

Once while emergency braking from 20 km/h the tester was able to decelerate at 0.31 g (the highest value achieved); for an unknown reason the left wheel stopped braking causing a 90⁰ spin. The tester had to put feet down to keep from falling.

Use of electric kickstands

General use of the Nino® led to the discovery of situations that pose concern when lowering the electric kickstands (Figure 3 ). First, we tested a situation when the rear kickstand comes down on an unforeseen object. This could cause the Nino® to tilt and accelerate uncontrollably, sometimes spinning, and, in our judgement, potentially causing the user to fall off the device. We also tested two related scenarios.

A test was conducted where blocks were placed under where the rear kickstand would land on the ground. No mass was added to the device. The block thickness was increased by increments of 6.35mm (0.25”). A rope was used to prevent damage to the device from falling over, which would often result from the spin induced by this scenario. All tests were filmed so the response could be analyzed later and rated for safety using either 0 (safe),  1 (potentially unsafe: the behaviour did not cause an unsafe situation but in a less controlled environment had potential to. This included the motors starting to accelerate or the device starting to spin but the standby mode was initiated before these behaviours could present any danger), 2 (unsafe: the device begins an uncontrolled fast spin or user control is lost) or Too Risky (situation was determined to be too unsafe for user testing). The safety rating was done independently by three people and the results were combined (Table 2 ).

Table 2. Safety of Nino behaviour with varying block sizes

Block Height (mm)	Safety Rating
	1 Loose Block no Mass	2 Loose Block with Mass	3 Fixed Block with Mass
6.35	0	0	0
12.7	0	0	0
19.1	0	0	0
25.4	1	1	1
31.8	1	2	1
38.1	1	Too Risky	1
44.5	2	Too Risky	1
50.8	2	Too Risky	1
57.2	2	Too Risky	1
63.5	2	Too Risky	1
69.9	2	Too Risky	1
76.2	2	Too Risky	1

The second test was conducted exactly as the first but with a 73 kg user operating the device. The mass made the test more similar to a real situation and reduced the block height that will cause an unsafe result (Table 2 ).

The third test used the same user and blocks but this time the blocks were fixed to the floor to more closely simulate another type of real situation (Table 2 ). The behaviour was never classified to be unsafe, but above 25 mm the Nino® ended up in its active state with the kickstands partially down which was rated with a one.

DISCUSSION

The unique operating characteristics of the Nino® can be attributed to the finding that the greater the speed the greater the deceleration force produced during braking. When moving faster a larger tilt angle is possible without tipping which allows for a higher deceleration; at slower speeds if that same tilt were achieved the Nino® would tip over backwards. More importantly, the braking distances of the Nino® ranged from 1.5 to 6.1 m depending on speed. At higher speeds, our measurement was higher than the company’s reported braking distance of 4 m. The measured and reported braking distances are also much higher than those of conventional PWCs, which have braking distances of about 1.3 m from speeds of 6 to 12 km/h [9]. These PWCs also have higher deceleration forces of up to 0.35 g [9]. The braking distance of the Nino® may be a concern for end-users and prescribers, especially depending on the use cases. However, one benefit to the way the Nino® stops is that the user is actually pushed into the seat by the tilt, decreasing the potential to be pitched out in a hard-braking event.

Since the Nino® is based on Segway® technology, it may be instructive to compare the safety issues of the conventional Segway® when used by able-bodied people. A recent systematic review of the literature concluded that Segway® “use is associated with a wide range of injuries. Many of these injuries require hospital admission and surgical intervention, incurring significant morbidity and high costs” [10]. YouTube searches for “Segway® crashes” find many videos of crashes, often related to sudden stopping. The higher CoG of standing on the Segway® may allow a user to shift their CoG quicker than when seated on the Nino®, leading to quicker stopping but also less safe conditions. However, as experienced while emergency braking at 0.31 g it is still possible to lose control while braking hard on the Nino®. Other Segway® crashes are related to turning since the higher CoG has the potential to throw the user off during a sharp turn. This may happen less on the Nino® because of the lowered CoG and seated position, which is more stable during cornering. Of course, any advantages of being seated are negated during a crash since a disabled user cannot jump from the device or use their legs and feet to stabilize the fall.

Our results showed that it is very important to check the floor and surroundings before engaging the kickstands and entering standby mode on the Nino®. The most dangerous situation happened when the kickstands interacted with a movable object (e.g. a purse or book left on the floor) with a height greater than 25 mm. A fixed difference in height (e.g. a height difference in floor surfaces or a raised threshold) was safer. However, the active state with kickstands partially down can cause serious hazards. Anecdotal evidence showed that if the user was unaware that the Nino® was still active and leaned to accelerate, the kickstands would dig in causing the device to spin. Any situation where the height difference of the two kickstands is above 25 mm is potentially very dangerous, sometimes causing quick accelerations and/or uncontrolled spins. These occurrences are also very hard to recover from – the tester often had to stand up completely and lift Nino® off the block. He also sometimes had to jump out of the device to safety, much like Segway® users seen in YouTube crash videos. For many people with disabilities, this may not be possible.

With the Nino it is very important to make sure the kickstands contact the ground simultaneously. A hazardous situation occurs when lowering the kickstands facing up or down a slope. This is specifically warned against in the user manual but this may still happen during regular use. The side faced uphill touches first and tilts the Nino® downhill, causing the Nino® to accelerate. The downhill kickstand digs in, again causing the Nino® to spin quickly. This reaction has a very large potential to pitch the user off or to tip the entire device. The solution is to turn the Nino® 90° to the hill and then extend the kickstands, allowing them to contact the ground simultaneously. While this provides a safe solution, our testing highlighted the importance of user awareness in these situations.

Lowering the kickstands while moving also presented potential hazards. No formal testing was performed but anecdotal evidence indicates that, similar to the above situations, one kickstand digs in and the whole device pivots around it, potentially pitching the user or tipping the device. The kickstands can also be retracted 15 seconds after the Nino® is powered off, making the device fall over. Similarly, the Nino® can be turned off without the kickstands engaged, also resulting in device falling. These situations are warned against in the manual, but an improved design is recommended, including improved interaction between the kickstands and the system when entering standby mode.

A counterintuitive situation occurs when the user encounters the potential of running into an object. A user with leg function will intuitively extend their legs forward or onto the ground to try and stop, which has the negative affect of tilting the Nino® forward causing acceleration into the object. A similar scenario occurs when reaching forward to activate a cross-walk button. The downslope of the curb cut tends to cause the Nino® to roll forward. This, compounded with the forward CoG shift make unintentionally entering traffic a real possibility. Training may mitigate some of these issues.

We also experienced the possibility of pulling the tiller steering column out during use. It is meant to be removed for transfers, but does not lock in place during movement. Pulling up will easily dislodge it and the user will be unable to steer, but the Nino® will still balance effectively. It is difficult to put the column back in without deploying the kickstands, as looking forward and down causes a CoG shift and the Nino® accelerates forward.

FUTURE RESEARCH

Kickstand block scenarios with a safety rating of one (potentially unsafe), would be interesting to expand upon to get a more detailed understanding of where the safe or unsafe line is. This would help to understand other factors found in less controlled environments that contribute to the Nino’s behaviour other than only block height.

Our work has revealed potential hazardous situations that may occur during typical use of the Nino® wheeled mobility device. Given our limited testing, we envision other hazards exist, some of which are explained in the Nino® user manual, and some which are like those known to occur when using a Segway® [8]. For instance, use on slippery surfaces such as grass or wet surfaces may present potential hazards and should be explored further.

During our MWC user evaluation (currently under journal review), users expressed worries about how the Nino® would react if the user had a spasm [7]. The device is sensitive to CoG changes, this situation may become very problematic and should be explored in the future.

While reviewing videos of YouTube Segway® crashes, a common situation was when one wheel accidently hit a solid object, causing a fall. This would eventually occur with general use of the Nino® or any wheelchair for that matter. The issue being the inability of the user to compensate during a fall. Nino® behaviour and user reaction during these situations would be very interesting to investigate to develop fall mitigation measures.

CONCLUSION

The Nino® and other self-balancing devices represent exciting new technologies that may, in the future, offer improved mobility for people with disabilities. The Nino® with its compact design, highlights the user and not the device. However, safety concerns and potentially hazardous use scenarios need to be fully understood. The solutions may be design changes, user training, and/or simply further understanding of situations to avoid completely. With any new technology, especially one that is so linked to the user’s wellbeing, appropriate steps must be taken to ensure the device safety before the full range of new benefits can be realized.