The start-ups using artificial intelligence to solve everyday tasks

The start-ups using artificial intelligence to solve everyday tasks

Dr Matthew Roberts - Senior Consultant, Data Exploitation

By: Matthew Roberts
Senior Consultant, Data Exploitation

5th July 2017

Home » Design

I recently attended the inaugural Cambridge Wireless Artificial Intelligence & Mobility Conference. The event focussed on artificial intelligence (AI), the business use cases enabled by AI, innovative start-up companies, and how start-up companies can gain funding. Unlike the technical conferences that I am used to attending, this event was much more about the business-side of AI.

Like many engineers, I usually like to look at the technical aspects of things, but this event gave me a different, and somewhat refreshing, perspective on the use of AI. I enjoy hearing about how companies, like DeepMind, are using AI to play video games and diagnose medical conditions, but perhaps I don’t pay enough attention to the companies that are using AI to solve everyday tasks. The Cambridge-based event gave start-ups the opportunity to talk and exhibit and gave people like me the chance to learn more about them.

You have probably heard of the driverless car technology being developed by organisations like Google and Uber, but what you might not know about are the driverless cars in the UK. Three driverless car projects were awarded funding by the UK government, and members of the public were given the opportunity ride in driverless cars.

Oxbotica, an Oxford University spinout, was involved in two of the projects. Oxbotica’s Selenium software formed the brains of the vehicles used in both projects. The software almost certainly uses AI to perform two key tasks: understanding the wealth of sensor data that is used to observe the car’s environment and controlling the car.

Another company that is working on self-driving cars is FiveAI. At the event, Stan Boland, CEO of FiveAI, spoke of how FiveAI is aiming to become a customer to large organisations instead of a supplier. FiveAI intends to do this by competing with the likes of Uber, but with self-driving cars. The company is currently part of a consortium that plans to test such cars on public roads in London, and AI will be a key part of making that a success.

Hoxton Analytics is using AI to solve a completely different kind of perception task. It is using cameras combined with AI to measure footfall. The cameras are mounted at ground level in order to avoid privacy concerns. Not only can it be used by shops to determine how many people it can attract, but it can also be used to infer the types of shoppers. This information can be used to help determine which demographics are being lured into shops and at what times. Solving such a task manually can be very labour-intensive.

Another example of the use of AI to solve everyday tasks is the 3D sensor that has been created by Titan Reality. Titan Reality’s sensor can be used in a wide variety of perception and control tasks, from sorting objects to pouring the correct drink based on what kind of glass is placed on the sensor.

This is just a tiny set of examples of where small companies have embraced AI to provide high-tech solutions to everyday tasks that would traditionally be performed by people. It is not just large companies like Google and Netflix that are using AI to make a big impact.

Save

Save

Save

Save

Save

Save

I recently attended the inaugural Cambridge Wireless Artificial Intelligence & Mobility Conference. The event focussed on artificial intelligence (AI), the business use cases enabled by AI, innovative start-up companies, and how start-up companies can gain funding. Unlike the technical conferences that I am used to attending, this event was much more about the business-side of AI.

Like many engineers, I usually like to look at the technical aspects of things, but this event gave me a different, and somewhat refreshing, perspective on the use of AI. I enjoy hearing about how companies, like DeepMind, are using AI to play video games and diagnose medical conditions, but perhaps I don’t pay enough attention to the companies that are using AI to solve everyday tasks. The Cambridge-based event gave start-ups the opportunity to talk and exhibit and gave people like me the chance to learn more about them.

You have probably heard of the driverless car technology being developed by organisations like Google and Uber, but what you might not know about are the driverless cars in the UK. Three driverless car projects were awarded funding by the UK government, and members of the public were given the opportunity ride in driverless cars.

Oxbotica, an Oxford University spinout, was involved in two of the projects. Oxbotica’s Selenium software formed the brains of the vehicles used in both projects. The software almost certainly uses AI to perform two key tasks: understanding the wealth of sensor data that is used to observe the car’s environment and controlling the car.

Another company that is working on self-driving cars is FiveAI. At the event, Stan Boland, CEO of FiveAI, spoke of how FiveAI is aiming to become a customer to large organisations instead of a supplier. FiveAI intends to do this by competing with the likes of Uber, but with self-driving cars. The company is currently part of a consortium that plans to test such cars on public roads in London, and AI will be a key part of making that a success.

Hoxton Analytics is using AI to solve a completely different kind of perception task. It is using cameras combined with AI to measure footfall. The cameras are mounted at ground level in order to avoid privacy concerns. Not only can it be used by shops to determine how many people it can attract, but it can also be used to infer the types of shoppers. This information can be used to help determine which demographics are being lured into shops and at what times. Solving such a task manually can be very labour-intensive.

Another example of the use of AI to solve everyday tasks is the 3D sensor that has been created by Titan Reality. Titan Reality’s sensor can be used in a wide variety of perception and control tasks, from sorting objects to pouring the correct drink based on what kind of glass is placed on the sensor.

This is just a tiny set of examples of where small companies have embraced AI to provide high-tech solutions to everyday tasks that would traditionally be performed by people. It is not just large companies like Google and Netflix that are using AI to make a big impact.

Save

Save

Save

Save

Save

Save

5 things to think about when building a combative robot

5 things to think about when building a combative robot

Rob Karplinski - Project Engineer, Embedded Systems

By: Rob Karpinski
Project Engineer, Embedded Systems

21st June 2017

Home » Design

Advances in computer science and engineering have brought in great change for robotics over recent years. From four-armed marimba robots being capable of composing original music to self-balancing quadrupedal robots capable of negotiating rough outdoor terrain, robotics has come a long way since the Roomba, one of the first robot vacuum cleaners.

However, if there is one thing better than putting together robots, it’s taking them apart again with other robots. With that in mind, and with the arrival of Series 9 of Robot Wars earlier in the year, here are five things to think about if you’re looking to build a combative robot for the hit TV show.

1. Get a good team together

Building a heavyweight, 110kg robot is hard and, talking from experience, there is a steep learning curve. It’s better to share this with a team of 3 – 4 people than taking it all on yourself. You may find it helpful to divide the responsibilities into certain roles like, design, construction, and buying. Alternatively, you can split the roles by system: drive, weapon, and chassis for example.

2. Choosing a chassis & materials

Most robots use a special type of steel, called Hardox, for armour. This wear-resistance steel is often used in digger buckets and is as strong as Titanium. However, unlike Titanium, it will bend rather than crack.

The limiting factor for robot size is the weight. A bigger robot will have to have thinner, lighter armour than you would have for a smaller robot. Keep in mind that an ideal robot is as small as the weapon and engine will let it be.

There is a growing trend to using Computer Aided Design (CAD) packages to design your robot. With many professional quality packages being offered for free by companies, this allows you to send design files directly to the metal working companies. They will then cut and shape parts of your robot for you, for a cost. This method is often quite expensive, however, will allow you to build a robot if you don’t have access to a fully-equipped workshop. Local makerspace/hackerspace communities may also have access to suitable workshops with metal working tools.

A good substitute for metal is a plastic called high-density polyethylene (HDPE). Using this resilient plastic can cut costs down and can be shaped with easily accessible woodworking tools. However, the thicknesses required to build a suitably robust robot leaves little to no weight advantage over using Hardox.

3. Have an exciting weapon

Robot Wars is an entertainment show, therefore the live audience and the viewers at home, will want to see destruction. Being a combative robot, it is common advice to plan the weapon first and then design the rest of your robot around it. Before getting started on the weapon, familiarise yourself with the Robot War rules first as they are quite specific. This is mainly due to safety but also to keep it exciting for the audience. Also make sure that your robot is robust enough to handle a hit from your own weapon.

Common weapon types are spinners, flippers and thwackbots. Spinners will do the most damage, both to your opponent and your own robot. Flippers are not as devastating but can be very impressive and effective if used tactically. It’s always entertaining to watch a robot sail a couple of meters into the air! Thwackbots are the simplest to make, but quite hard to make exciting.

4. Choosing motors and steering systems

In Robot Wars, if your robot doesn’t move for 10 seconds it will be counted out and you lose the match! However, if your weapon stops working, you can still dodge and push your opponent into the house robots or the dreaded pit.

Most robots use a 2 wheel tank-style steering system, however, 4 or more wheels allow for increased grip on the arena floor. Very few robots use tracks as they’re very tricky to get to work reliably.

The Bosch 750 brushed motor is a fan-favourite among the Robot Wars community as these have a good power-to-weight ratio. However, they’ve been discontinued now, so finding replacements and spares may be difficult but, because of its popularity, not impossible.

800W scooter motors available from China can make for good alternatives. The quality is not as great but they are more affordable and easily available. Regardless, having a brushed motor is recommended for most teams as they don’t require pricey speed controllers and are easy to set up. Brushless motors have a much better power to weight ratio but the controllers can be complicated so very few teams use these. Tune your robot for acceleration over speed as the secret to winning is quickly positioning your robot into a good spot to avoid the opponent’s weapon whilst attacking with your own.

5. Be an interesting robot

As a final note, have fun with it. Robots with amusing and interesting themes are more likely to be accepted than boring wedges or tin cans. LED’s are cheap and don’t require much power to light up your robot. Robots with great paint schemes, team costumes, and novel backstories can become hugely popular. Get the audience to buy into your robot with a catchy name that is easy to chant. This can be hard to get right but is worth the time spent. You can always run suggestions and ideas past friends to see what they think.

Save

Save

Save

Save

Save

Save

Advances in computer science and engineering have brought in great change for robotics over recent years. From four-armed marimba robots being capable of composing original music to self-balancing quadrupedal robots capable of negotiating rough outdoor terrain, robotics has come a long way since the Roomba, one of the first robot vacuum cleaners.


However, if there is one thing better than putting together robots, it’s taking them apart again with other robots. With that in mind, and with the arrival of Series 9 of Robot Wars earlier in the year, here are five things to think about if you’re looking to build a combative robot for the hit TV show.

1. Get a good team together

Building a heavyweight, 110kg robot is hard and, talking from experience, there is a steep learning curve. It’s better to share this with a team of 3 – 4 people than taking it all on yourself. You may find it helpful to divide the responsibilities into certain roles like, design, construction, and buying. Alternatively, you can split the roles by system: drive, weapon, and chassis for example.

2. Choosing a chassis & materials

Most robots use a special type of steel, called Hardox, for armour. This wear-resistance steel is often used in digger buckets and is as strong as Titanium. However, unlike Titanium, it will bend rather than crack.

The limiting factor for robot size is the weight. A bigger robot will have to have thinner, lighter armour than you would have for a smaller robot. Keep in mind that an ideal robot is as small as the weapon and engine will let it be.

There is a growing trend to using Computer Aided Design (CAD) packages to design your robot. With many professional quality packages being offered for free by companies, this allows you to send design files directly to the metal working companies. They will then cut and shape parts of your robot for you, for a cost. This method is often quite expensive, however, will allow you to build a robot if you don’t have access to a fully-equipped workshop. Local makerspace/hackerspace communities may also have access to suitable workshops with metal working tools.

A good substitute for metal is a plastic called high-density polyethylene (HDPE). Using this resilient plastic can cut costs down and can be shaped with easily accessible woodworking tools. However, the thicknesses required to build a suitably robust robot leaves little to no weight advantage over using Hardox.

3. Have an exciting weapon

Robot Wars is an entertainment show, therefore the live audience and the viewers at home, will want to see destruction. Being a combative robot, it is common advice to plan the weapon first and then design the rest of your robot around it. Before getting started on the weapon, familiarise yourself with the Robot War rules first as they are quite specific. This is mainly due to safety but also to keep it exciting for the audience. Also make sure that your robot is robust enough to handle a hit from your own weapon.

Common weapon types are spinners, flippers and thwackbots. Spinners will do the most damage, both to your opponent and your own robot. Flippers are not as devastating but can be very impressive and effective if used tactically. It’s always entertaining to watch a robot sail a couple of meters into the air! Thwackbots are the simplest to make, but quite hard to make exciting.

4. Choosing motors and steering systems

In Robot Wars, if your robot doesn’t move for 10 seconds it will be counted out and you lose the match! However, if your weapon stops working, you can still dodge and push your opponent into the house robots or the dreaded pit.

Most robots use a 2 wheel tank-style steering system, however, 4 or more wheels allow for increased grip on the arena floor. Very few robots use tracks as they’re very tricky to get to work reliably.

The Bosch 750 brushed motor is a fan-favourite among the Robot Wars community as these have a good power-to-weight ratio. However, they’ve been discontinued now, so finding replacements and spares may be difficult but, because of its popularity, not impossible.

800W scooter motors available from China can make for good alternatives. The quality is not as great but they are more affordable and easily available. Regardless, having a brushed motor is recommended for most teams as they don’t require pricey speed controllers and are easy to set up. Brushless motors have a much better power to weight ratio but the controllers can be complicated so very few teams use these. Tune your robot for acceleration over speed as the secret to winning is quickly positioning your robot into a good spot to avoid the opponent’s weapon whilst attacking with your own.

5. Be an interesting robot

As a final note, have fun with it. Robots with amusing and interesting themes are more likely to be accepted than boring wedges or tin cans. LED’s are cheap and don’t require much power to light up your robot. Robots with great paint schemes, team costumes, and novel backstories can become hugely popular. Get the audience to buy into your robot with a catchy name that is easy to chant. This can be hard to get right but is worth the time spent. You can always run suggestions and ideas past friends to see what they think.

Save

Save

Save

Save

Save

Save

Save

Save

Save

An insight from a PCB Design Apprentice

An insight from a PCB Design Apprentice

Elliot Langran - PCB Design Apprentice

By: Elliot Langran
PCB Design Apprentice

26th April 2017

Home » Design

Hello, I’m Elliot Langran and I am the PCB Design Apprentice here at Plextek, I started in September and here is an insight from my perspective of life as an apprentice.

I generally get into the office between 9 – 9:30 am, being a young person, I’m extremely grateful for our flexible working hours. It does mean staying later than most, but at least I can claim I’m running the show later in the day, if only for an hour or so. I start the day catching up with the team and of course grabbing a cup of tea to get me kick-started. The first thing I do is get everything set up including the Computer Aided Design (CAD) tools, and PADS layout, the Printed Circuit Board (PCB) design software. I’d be lost without these.

PCBDesign4-500Being a PCB Design Apprentice is really varied and my average day can be spent doing a range of different things, generally speaking, no two tasks are the same. When I started my apprenticeship, I luckily wasn’t expected to know how to design a 12 layered PCB.

However, excitingly, I am now on my way to gaining the relevant skills. This makes a 12 layered PCB less of a daunting idea as each day I am developing and learning.

One thing that has become apparent to me is that each PCB designer has their own style as each designer lays out their board differently (it’s nice to know that I will develop a PCB signature). Understanding and being exposed to a range of layout options and developing multiple skill sets is vital for me in crafting my own style further.

PCBDesign1-500What I enjoy about the job is the creativity and it’s one of the first principles that I was taught here: never copy anybody else’s design, and develop your own style and way of working instead. This is helped by the constant support and guidance from my mentors. They are always advising and guiding me on where improvements can be made and are key people in my development as an apprentice.

I’ve been learning a lot by looking at the other designers’ work from time to time and noting their style of approaching layouts for example. They can work on as many as 16 layers of PCB board, it’s quite crazy actually. I am currently able to work up to 4 layers and the tracking gets more complicated each time you go up a layer pair. This is made even more challenging if you have less board space to work on.

I also spend some of my time working on mechanical drawings, which really brings extended variety to my work day! These drawings are needed often during projects, especially for cable assemblies.

PCBSo there is a lot of variety in my job; PCB design, schematics and sometimes mechanical drawings, not to mention college work as well. My line manager and mentor are always really supportive of my college work and both think it is important to get done whenever there is a quiet moment. This has been vital to my success in my apprenticeship studies and is greatly appreciated.

The organisation is always showing great enthusiasm towards the apprenticeship scheme and my personal development. This has been recognised by my college, Cambridge Regional College, at this year’s Apprenticeship Awards as we won “Engineering Employer of the Year Award”. I look forward to developing here during my apprenticeship and the adventures and interesting projects that await me.

Hello, I’m Elliot Langran and I am the PCB Design Apprentice here at Plextek, I started in September and here is an insight from my perspective of life as an apprentice.

I generally get into the office between 9 – 9:30 am, being a young person, I’m extremely grateful for our flexible working hours. It does mean staying later than most, but at least I can claim I’m running the show later in the day, if only for an hour or so. I start the day catching up with the team and of course grabbing a cup of tea to get me kick-started. I first get everything set up including the Computer Aided Design (CAD) tools, and PADS layout, the Printed Circuit Board (PCB) design software. I’d be lost without these.

PCBDesign4-500Being a PCB Design Apprentice is really varied and my average day can be spent doing a range of different things, generally speaking, no two tasks are the same. When I started my apprenticeship, I luckily wasn’t expected to know how to design a 12 layered PCB.

However, excitingly, I am now on my way to gaining the relevant skills. This makes a 12 layered PCB less of a daunting idea as each day I am developing and learning.

One thing that has become apparent to me is that each PCB designer has their own style as each designer lays out their board differently (it’s nice to know that I will develop a PCB signature). Understanding and being exposed to a range of layout options and developing multiple skill sets is vital for me in crafting my own style further.

PCBDesign1-500What I enjoy about the job is the creativity and it’s one of the first principles that I was taught here: never copy anybody else’s design, and develop your own style and way of working instead. This is helped by the constant support and guidance from my mentors. They are always advising and guiding me on where improvements can be made and are key people in my development as an apprentice.

I’ve been learning a lot by looking at the other designers’ work from time to time and noting their style of approaching layouts for example. They can work on as many as 16 layers of PCB board, it’s quite crazy actually. I am currently able to work up to 4 layers and the tracking gets more complicated each time you go up a layer pair. This is made even more challenging if you have less board space to work on.

I also spend some of my time working on mechanical drawings, which really brings extended variety to my work day! These drawings are needed often during projects, especially for cable assemblies.

PCBSo there is a lot of variety in my job; PCB design, schematics and sometimes mechanical drawings, not to mention college work as well. My line manager and mentor are always really supportive of my college work and both think it is important to get done whenever there is a quiet moment. This has been vital to my success in my apprenticeship studies and is greatly appreciated.

The organisation is always showing great enthusiasm towards the apprenticeship scheme and my personal development. This has been recognised by my college, Cambridge Regional College, at this year’s Apprenticeship Awards as we won “Engineering Employer of the Year Award”. I look forward to developing here during my apprenticeship and the adventures and interesting projects that await me.

Save