The disruptive technology hype is abuzz with innovations such as the driverless or autonomous cars, drones, artificial intelligence among others. All these innovations are exciting, and I believe it is a question of time before we see these in Zimbabwe. Already, if you go to certain events like weddings or church meetings, you see drone cameras being used. In a lot of developed countries, autonomous cars are being deployed and there is a race among manufacturers such as Toyota, BMW, Mercedes Benz and Tesla to greatly enhance the technology for safety. Surely, Zimbabwe cannot be left out. We need to adapt. So let us look at the advent of the autonomous cars on our roads in Harare, or even countrywide.
guest column: Tororiro Isaac Chaza
Let us first try to unpack the technology required for an autonomous vehicle to safely navigate the roads. Wikipedia describes the technology thus; “a self-driving car, also known as a robot car, autonomous car, or driverless car, is a vehicle that is capable of sensing its environment and moving with little or no human input”. In order for this vehicle to be driverless it needs to work with a number of sensors “such as radar, computer vision, Lidar, sonar, GPS, odometry and inertial measurement units. Advanced control systems interpret sensory information to identify appropriate navigation paths, as well as obstacles and relevant signage.”
As a technologist, I am tempted to proceed with the article without further explanation of these technologies, but alas, I would lose a significant number of readers. So, I will attempt to explain some of the technologies in very simple terms. Stay with me. You may be familiar with the name “radar”, which was an acronym for “Radio Detection and Ranging’” technology for detecting the position of airplanes, ships among others.
It is no longer an acronym, but has been incorporated into the English dictionary as a word. Another technology is “Sonar”, standing for “Sound Navigation and Ranging”, which uses sound to detect other objects in the vicinity, and their distance. “Lidar” is similarly “Light Detection and Ranging”, or alternatively “Light Imaging Detection and Ranging”.
“Lidar” uses a sensor, which uses laser light beamed towards a target object, and measures the time it takes to receive the laser light reflected back towards the sensor. The sensor calculates the distance to the target by measuring the time it takes to receive the reflected light. I will make it even simpler. Take a tennis ball and stand one metre from a wall. Throw the tennis ball and count very fast one, two, three… till you catch it when it bounces back. You will probably count only up to two or three. Now move back 10 metres and throw the tennis ball to the wall again counting one, two, three …. Until you catch the bounced ball again. You will probably have counted up to seven. In other words, the longer the distance from the wall, the more the count illustrating the longer time it takes for the ball to bounce back. In the same way as a tennis ball, laser light can be emitted in pulses towards an object and the further the object, the longer the light will reflect back. The only difference being that the Lidar is a million times more precise in measuring than your counting. If you still cannot understand, ask your Grade Seven child. Lidar can scan an object in a three dimensional (3D) perspective. Combined with computer technology, global positioning system (GPS), Sonar and other sensor technologies, it is possible for a driverless car to navigate roads by mimicking what a human driver would see, perhaps even surpass the human being. So let’s make Zimbabwe great again by getting ready for the autonomous car on our roads. What do we need to prepare for the advent of the driverless car?
It will be difficult to just import a driverless car into Zimbabwe without a massive effort to customise it with bespoke software and sensor technology to suit the local conditions.
Firstly, the navigation sensors and software would have to be enhanced to become pothole sensitive. I must admit there has been some improvement on some roads as they have been patched. For instance at one time the potholes on Arcturus Road were visible on Google maps. Now, they have disappeared. However, there is still too much work to do to improve the roads. Normally, the lidar is put on top of the car, but in Zimbabwe, one would also need an under-chassis lidar for spotting potholes. But will the driverless car be able to avoid the pothole fast enough? Alas, I do not think so. And what will Lidar make of the self-employed road maintenance personnel who are seen on our road who slowly, and very slowly repair potholes using bricks for a compassionate fee? Times are hard.
Secondly, there has to be customisation on the correct reading of traffic lights. In Harare, when a traffic light goes from “red” to “blank”, it means “go” because the light bulb for the green light is dead. Now, that is the rule on most intersections, but it becomes complicated on intersections with the “filter arrow”. The driverless car would have to be programmed to know which traffic lights go from “red” to “blank”, meaning “go” and/or you can also “filter” right. There are also some intersections where it does not matter what the traffic light is indicating, just go and squeeze in into the madness at your own peril.
At times there are gracious ‘street kids,’ more like adults though, who immediately become traffic wardens. They do an excellent job of trying to direct and clear the traffic gridlock at the intersections. Now, the driverless car would have to understand that the “street kid” in front who is not wearing any safety clothing such as that of police, and waving his hands is actually in charge of traffic control.
But if any accident were to happen because the “street kid” directed the driverless car to proceed without giving way to the car on the right, the “street kid’s” authority is rendered null and void as a defence. The normal rule of ‘give way to the vehicle on your right’ never seems to be working at all in Harare. The rule is “watch out”.
Thirdly, the autonomous car would have to be programmed to understand certain roads, especially to and from high-density suburbs (HDS), which have become very high-density in terms of vehicular traffic. The rule is that in the morning when most traffic is going from the HDS towards the town centre and you are going in the opposite direction, you do not have the right of way in your own lane. Just give way to the on-coming traffic which is in your lane. It is rightfully your lane, but if you do not get out of the way, you may cause a serious accident, which may be fatal to you and the kombi-full of innocent passengers. Even if you are not ‘killed’ I wonder what the insurance company will say as the kombi that rammed into you would most likely not be insured.
Fourthly, the intersections with a single ‘right turn’ lane are no longer the rule. One can make an additional lane from the outer lane with impunity. No one is there to police the violation. The autonomous car will have to be programmed to observe such a violation. On top of all that, there seems to be a dangerous ‘school run’ trend of school mums or dads talking on the cellphone and whatsApping, while driving, especially when turning, with a child sitting in the front seat.
Fifthly, the autonomous car’s lidar would have to scan and identify the m’shikashika (informal, even illegal taxis) vehicles and proceed with care or completely avoid being near these. So lidar would have to identify a Fun Cargo, Honda Fit, Toyota Wish or any such vehicle, and quickly decide if this is a m’shikashika or a decent law-abiding motorist. It is difficult to tell as even law-abiding motorists are tending to drive like the m’shikashika. It could also be a case of fraudulently obtained driving licences or no licences at all.
Sixthly, lidar must understand that even when the traffic light turns green for ‘go,’ don’t go yet. There is a very high probability (99,999%) that the car coming from the right, or left, at high speed against a red robot is not going to stop at all.
Seventhly, lidar must be programmed to read road signs. “What road signs?” you are going to ask. The ones that are collapsed by the road side. They are legal. You will be fined for not observing the “Give Away” sign, correction, the “Give Way” hidden in the maize field.
Eighthly, lidar must not be confused by recognising humans who travel at 80kph on our roads. Those are “hwindis” or touts on the back of a speeding kombi. I perceive this is a very dangerous stunt. If a “hwindi” were to fall at 80kph and the autonomous car is behind the kombi, there would be no time to stop. I wonder how they get away with it when I was always fined for an original German manufacturer’s fire extinguisher because it did not have a Standards Association of Zimbabwe stamp, or the manufacturer’s original spare “biscuit” wheel.
When it comes to long-distance driving, the autonomous car must be programmed to resist going on the road no matter what. But in case the regulatory authority insists that the autonomous car service provider must also cater for long distances, there is a whole new set of rules that need to be programmed for lidar to observe.
Chief among these rules is no night-time driving under any circumstances. Although lidar does not rely on sunlight to sense objects, it is still dangerous to drive at night for a number of reasons, some of them covered below.
Firstly, the autonomous car has to be programmed to recognise a “gonyet”. Many rules apply to “gonyet” caution. The autonomous car must be programmed to slow down very much when there is a “gonyet” coming from the opposite direction, most likely above the 80kph speed limit, and encroaching into the autonomous car’s lane. A “gonyet” is an eighteen wheeler truck or in Germany it is called a “sattelschlepper“.
Never overtake a “gonyet” when it is going downhill, as it gathers enormous momentum. Never overtake or drive in the opposite direction of an on-coming “gonyet” on a narrow bridge. I consider any bridge on a standard Zimbabwean highway to be narrow. “Gonyets” are notorious for fatally side-swiping smaller cars. Never overtake a “gonyet”, or any car for that matter, on a bend, a rise or when the road is marked by a continuous white line. Never overtake more than one “gonyet” at a time and do not follow behind any fool doing so.
Secondly, when overtaking, lidar must be aware of hitting side-road potholes at high speed. The Harare-Beitbridge Highway comes to mind. However this may not be a necessary caution since the road will be dualised by 2022 for $2,7 billion. This is a mega infrastructure project promised since 2003.
Thirdly, lidar must recognise animals with erratic behaviour such as cows, donkeys, goats, and baboons. Most farms have no fences to keep domestic animals off the roads. I have been a victim of this delinquency. In some areas there are even elephants crossing the road.
Fourthly, lidar must be programmed to anticipate police roadblocks at certain points. For instance, the speed limit is 120kph on open roads, but when approaching a built-up area such as a small town this is reduced to 80kph and then 60kph, and vice versa when exiting the small town. The roadblocks are set up to catch you going at 80kph when you have passed the 60kph speed limit sign. This is all above board.
Lastly, the driverless car will need to “understand” the road map using GPS. The current level of GPS information on Google maps would have to be accurate as there are a number of roads without names or wrongly labelled on the GPS map. Even if the roads are correctly named, there needs to be further work on street addresses as the current Google map does not lead me to the right street address.
I believe we can make the driverless car work and make Zimbabwe great again. Perhaps one more observation, lidar must recognise the presidential motorcade. I do not know how that could be programmed.
Tororiro Isaac Chaza vice-chairperson of Zimbabwe Information and Communication Technologies, a division of the Zimbabwe Institution of Engineers.